2024-25 Research Opportunites: Abudayyeh - Lotter

Current Projects: 

Building a “Virtual Cell”

We are pioneering the development of a “Virtual Cell,” using AI to predict and control cellular behavior. By integrating high-content single-cell data with machine learning, we aim to create a foundational model of cellular biology capable of reprogramming cells, uncovering new mechanisms, and accelerating drug discovery. This innovative approach has the potential to solve complex challenges in genetic disease, cancer, and aging, and beyond. As part of this effort, we are collecting the largest single-cell dataset ever compiled to build the largest biology model ever built. 

New modalities for disease treatment and human augmentation

Combining pooled screening approaches, single cell readouts, and machine learning methods, we are screening for new classes of peptide and protein drugs to treat diverse diseases and augment the human condition (e.g. neuromodulation). We are applying this model in multiple contexts, including muscle aging, inflammatory responses, immune system rejuvenation, and neural circuit modulation. 

Aging therapeutics, rejuvenation, and life span extension

We are applying novel molecular tools and approaches our lab has developed over the past few years to develop molecular signatures of aging and rejuvenate diverse aged tissues, including hematopoietic stem cells, muscle, and the immune system. 

Large language models for directed evolution and protein engineering

We are building the largest and most advanced models for protein design and engineering using generative foundation models and groundbreaking few-shot learning approaches. By integrating protein language models with active learning, we can accelerate the evolution of proteins, achieving dramatic improvements in protein function with minimal experimental rounds. This work paves the way for unparalleled advancements in protein engineering with applications across antibody therapies, genome editing, delivery, diagnostics, and sustainability/climate change. 

Nucleic Acid Delivery

The efficient delivery of nucleic acids into cells beyond the liver is critical for developing new gene and cell therapies. Our lab is leveraging the natural biology of nanoparticles and protein engineering to develop programmable delivery solutions to target extra-hepatic tissues. 

New gene editing tools

We have multiple projects creating novel systems to perturb and modify DNA and RNA. We deploy rational engineering methods, machine learning, and natural enzyme discovery, we hope to develop tools to unlock new classes of gene and cell therapies. 

Contact: hello-abugoot [at] mit.edu (hello-abugoot[at]mit[dot]edu)  

Location: MIT and Longwood

Current Projects: 

Medical image segmentation

Developing and applying computational analysis to segment brain structures from available T1/T2/diffusion MRI images. 

Brain connectivity analysis

Finding relationships between structural and functional connectivity of the human brain and neurodegenerative disease. 

Code optimization

Optimizing existing code so they run faster on CPU and GPU.

Contact: iaganj [at] mgh.harvard.edu (iaganj[at]mgh[dot]harvard[dot]edu)

Lab Location: Charlestown

Current Projects: 

Enhanced Ultrasound Imaging through Machine Learning/Artificial Intelligence (ML/AI)

Leverages advanced ML and AI algorithms to analyze and interpret ultrasound data with greater precision and efficiency. By automating the detection of subtle patterns and features within ultrasound images, ML/AI can improve diagnostic accuracy, reduce human error, and enhance the overall quality of imaging. These technologies can be applied in real-time, allowing for the dynamic adjustment of imaging parameters to improve resolution and contrast. Potential applications include early detection of diseases, real-time monitoring during medical procedures, and personalized imaging that adapts based on patient-specific characteristics.

Ambient Sensing for Vital Sign Monitoring

Use of advanced, non-intrusive sensors to extract critical vital sign data—such as heart rate, respiration rate, and body temperature—from participants without direct physical contact. By integrating these sensors with sophisticated algorithms, the system can continuously monitor participants in a comfortable and unobtrusive manner, ensuring high data accuracy. The selection and validation of sensor candidates, alongside the algorithms, must adhere to clinical standards to ensure reliability and accuracy in medical environments. This method is particularly beneficial for long-term monitoring of healthy participants, early detection of abnormalities, and improving patient comfort in both clinical and home-care settings.

Contact: banthony [at] mit.edu (banthony[at]mit[dot]edu) | Sam Young, samyoung [at] mit.edu (samyoung[at]mit[dot]edu) 

Location: MIT

Please visit our website and contact us to learn more about our current projects. 

Contact: nartzi [at] mit.edu (nartzi[at]mit[dot]edu) 

Location: Cambridge

Current Projects: 

The use of AI in Orthopedic Surgery 

We use various AI methods to improve the quality of healthcare in orthopedic surgery diagnosis, treatment, prediction of outcomes, and developing data registries. We turn our AI algorithms into applications and decision-support tools for our providers. These projects include entrepreneurship opportunities as well. 

3D modeling and orthopedic device development 

We design novel devices for the treatment of orthopedic patients, including patient-specific surgical tools, rehabilitation devices, and orthoses, using novel 3D printers we have in the lab. 

Clinical Studies in Orthopedics

Clinical studies, including retrospective studies, prospective trials, and review articles, are being conducted in our lab on various topics in the realm of orthopedic practice. 

Novel Imaging methodologies 

We use novel imaging techniques, including portable handheld ultrasound devices, weight-bearing bilateral CT scans, novel needle arthroscopy, and portable thermography imaging, to improve the accuracy, speed, and access to diagnostic methods for our providers. We combine these technologies with computer-assisted and AI-assisted interpretation.

Contact: Sashkaniesfahani [at] mgh.harvard.edu (Sashkaniesfahani[at]mgh[dot]harvard[dot]edu) | Dr Atta Taseh, ataseh [at] mgh.harvard.edu (ataseh[at]mgh[dot]harvard[dot]edu)

Location: 158 Boston Post Road, Weston, MA 

Our lab applies develops and applies computational methods for analyzing cancer epigenomes to advance precision oncology. We have several projects that are well-suited to students with a strong background in R and/or python. 

Current Projects: 

Analysis of novel liquid biopsy datatypes

We are analyzing epigenetic features of cell-free DNA to learn about non-mutational mechanisms of treatment resistance in cancer. (eg: https://www.nature.com/articles/s41591-023-02605-z) 

Cistrome-wide associations studies

We are applying a method for understanding how GWAS variants influence cancer risk through effects on the epigenome. (eg: https://www.nature.com/articles/s41588-022-01168-y) 

Fragmentomics

We are developing methods to learn about epigenomic features of cancer from cell-free DNA fragmentation patterns in clinical cancer specimens. 

Predicting function of noncoding variants

We are applying machine learning approaches to large epigenomic datasets to predict the effects of non-coding variants on gene regulation

Contact: sylvan_baca [at] dfci.harvard.edu (sylvan_baca[at]dfci[dot]harvard[dot]edu) | Lauren Stone, Lauren_Stone [at] dfci.harvard.edu (Lauren_Stone[at]dfci[dot]harvard[dot]edu)

Location: Longwood

Current Projects: 

Engineering Immunity against HIV

Using AAV to deliver broadly neutralizing antibodies as a means of preventing or treating HIV infection. 

Development of polyclonal vectored immunoprophylaxis

Using AAV vectors to engineer the delivery of bi-specific antibodies with novel capabilities against HIV. 

Understanding AAV immunogenicity

Exploring the immunology of gene transfer in humans and developing approaches to minimize host response against it.

Contact: abalazs [at] mgh.harvard.edu (abalazs[at]mgh[dot]harvard[dot]edu)

Location: MIT / The Ragon Institute

Current Projects: 

Therapeutic base and prime editing in hematopoietic stem cells to correct, ameliorate, or enhance blood cell functions. Studies range from technology development to target discovery to preclinical validation to first-in-human proof-of-concept. 

Nucleotide-resolution functional genomics to identify novel mechanisms of hematopoiesis and targets for blood disorders. Pooled screens with single cell and in vivo readouts. 

Manipulation of hematopoietic stem cells to enable therapeutic genome editing without chemotherapy. 

Contact: daniel.bauer [at] childrens.harvard.edu (daniel[dot]bauer[at]childrens[dot]harvard[dot]edu)

Location: Longwood 

Current Projects: 

Preclinical and clinical radiation therapy and imaging with novel nanoparticles

We are developing new nanoparticles for combination imaging (e.g. MRI, x-ray) and radiation therapy applications. One nanoparticle is in Phase 2 clinical testing at our institution and others are currently being evaluated pre-clinically. Projects can include lab bench work, animal research, computational studies, and image analysis. 

Preclinical and clinical multi-energy imaging with novel detectors

We are engaged in the design, development, testing, and clinical evaluation of novel multi-layer flat-panel imagers for multi-spectral applications. Projects include Monte Carlo simulations, reconstruction algorithm design, and image analysis.

Contact: ross_berbeco [at] dfci.harvard.edu (ross_berbeco[at]dfci[dot]harvard[dot]edu)

Location: Longwood

Current Projects: 

A computational tumor model for Glioblastoma Multiforme (GBM)

To develop an innovative and complex approach to model the evolution of glioblastoma multiforme tumors. We will use a novel agent-based model to incorporate specific mechanisms specific to GBMs, helping treatment design and optimization. 

Personalizing Y90-microsphere treatment for transarterial radioembolization in liver cancer

We are working on a novel treatment planning system, with two main aspects to be further developed: (i) our current adult human liver phantoms models must be replaced by patient-specific data, and (ii) microspheres must be statistically distributed around the tumor and arteriovenous junctions in the normal tissue. Then, a dose calculation algorithm like Monte Carlo can be integrated into the MIDOS model. 

Modeling radiation-induced damage repair

How the DNA damage induced by radiation is repaired is a key factor in determining cell fate. We have recently developed a Monte Carlo-based model to simulate repair mechanisms according to a set of parameters. The student will be responsible for: collecting data from the literature showing DNA damage repair after controlled irradiations, finding the right values for the parameters in our model to reproduce observed experiments, and simulating the effects of different radiation types over several cell lines.

Contact: abertoletreina [at] mgh.harvard.edu (abertoletreina[at]mgh[dot]harvard[dot]edu)

Location: MGH Main Campus

Current Projects: 

We develop data acquisition and image reconstruction strategies that synergistically employ MR physics, cutting edge hardware, signal processing and deep learning algorithms to push the limits of spatial and temporal resolution for more efficient clinical and neuroscientific imaging. 

Contact: bbilgic [at] mgh.harvard.edu (bbilgic[at]mgh[dot]harvard[dot]edu)

Location: Charlestown

Current Projects: 

Transpinal Magnetic Stimulation

The project’s cornerstone is the integration of high-frequency neuromodulation, a method proven effective in invasive spinal cord stimulation therapies, into a non-invasive modality that targets the spinal structures non-invasively. By creating a compact, multilayer High-Frequency Trans-Spinal Magnetic Stimulation (HF-TSMS) system, we aim to modulate the neural pathways associated with pain transmission and management directly, offering a significant advantage over traditional surgical intervention methods. HF-TSMS seeks to mitigate the pain associated with sciatica, potentially reducing or eliminating the reliance on pharmacological treatments, which are often accompanied by significant side effects and the risk of addiction.

Metamaterial DBS and ECoG Electrodes for Magnetic Resonance Imaging

The proposed research aims to design, develop, and test novel NiTi using metamaterial technology and an absorbable ECoG electrode set for MRI-compatible Deep Brain Stimulation. The development of such novel technology could result in significant benefits for patients who suffer from some medically refractory pathological conditions such as Parkinson’s disease, epilepsy, and stroke. 

Simultaneous functional MRI and Micro-Magnetic Nervous System Stimulation

The proposed research aims to design, develop, and test a novel micro-magnetic stimulation system for next-generation nervous system stimulation, both in basic research and clinical applications. The development of such novel technology could result in significant benefits to basic neuroscience research as it will pair magnetic stimulators with Calcium channels optical sensors to measure onsite efficacy and fMRI for the significant network response.

Deep Brain Stimulation (DBS)

A neurosurgical intervention in which electrodes are implanted in the brain to stimulate specific target areas for the treatment of movement disorders and other conditions. However, there are concerns regarding MRI compatibility and the safety of current DBS implants, as the lead can act as an antenna, potentially causing substantial heat-related injuries. We are working on the fabrication of DBS leads with a novel design that ensures compatibility with MRI up to 3T.

Contact: Giorgio.Bonmassar [at] mgh.harvard.edu (Giorgio[dot]Bonmassar[at]mgh[dot]harvard[dot]edu) | Francesca Marturano, fmarturano [at] mgh.harvard.edu (fmarturano[at]mgh[dot]harvard[dot]edu)

Location: Charlestown

Current Projects: 

Self-supervised learning for pan-cancer foundation models

Self-supervised learning methodologies have driven the latest advances in machine learning over the last 2-3 years, but they tend to create "global" representations that consider aspects of the entire image, and hence information about clinically-significant abnormalities such as cancer is overwhelmed by non-clinically-significant features encoding normal anatomical variation. In this project, we will develop self-supervised methods using large weakly-labelled cancer imaging datasets to develop representations that focus on clinically-significant features in the hope of unlocking large clinical data archives to create foundation models for downstream imaging phenotype discovery. 

End-to-end deep learning for clinical radiology imaging

As deep learning has evolved, prior multi-step computational pipelines have gradually been subsumed into "end-to-end" trainable models that take "raw" pixel data and make high-level predictions of clinical value. However, in practice significant data curation, often using ad-hoc rules, is still required to present models with relevant, categorized pixel data. This is especially true in modalities such as CT and MRI, where an imaging study often consists of multiple different acquisitions with different acquisition parameters. In this project, we will develop methodologies to learn truly end-to-end, dynamic deep learning architectures for radiology that accept entire raw DICOM imaging studies and learn how to process and combine them to make predictions of interest.

Location: Charlestown

Contact: cbridge [at] mgh.harvard.edu (cbridge[at]mgh[dot]harvard[dot]edu) 

Current Projects: 

Development of PET probes for quantitative multimodal imaging of fibroproliferative diseases

We are developing novel probes that target molecular markers of fibrosis and applying these to questions of detection, prognosis, staging, and treatment monitoring in diseases like chronic kidney disease, chronic liver diseases, and heart failure. Opportunities in chemistry, imaging, modeling, disease models 

Development of novel radiotheranostics

We are designing molecules that target unique features of solid tumors. These molecules are radiolabeled with a positron emitting isotope for quantitative imaging and with a high energy beta or alpha emitting isotope to deliver radiotherapy to the tumor. Opportunities in chemistry, imaging, quantitative modeling, and disease models. 

PET-MR Imaging of pulmonary fibrosis

This project aims to apply PET-MR imaging to quantify molecular abnormalities in the lungs of idiopathic pulmonary fibrosis patients and determine if such measures can predict the pace of disease progression and determine whether the patient is responding to anti-fibrotic therapy.

Contact: pcaravan [at] mgh.harvard.edu (pcaravan[at]mgh[dot]harvard[dot]edu)

Location: Charlestown

Current Projects: 

Integrated predictions of clinical risk

Patients who carry rare variants in established disease genes (e.g. BRCA1, LDLR) may have increased risk for established clinical disorders. However, there are many other factors which can influence clinical risk, including polygenic risk, clinical risk factors, and behavioral/environmental/lifestyle risk factors. We develop integrated clinical risk models which draw on these sources of information. 

Approaches to assess variant functional effects

We develop integrated models to assess variant functional effects using population data, computational predictions of functional effect, and functional data from high-throughput CRISPR assays. These are useful for improving assessment of variant pathogenicity, but also for identifying new genes which may affect a trait or cause a disease. 

Developing clinical evidence to expedite variant classification

We design approaches to generate evidence which can be used in variant classification (e.g. determining whether a variant is classified by a clinical lab as pathogenic). These include developing evidence standards, calibrating the strength of evidence of pathogenicity for a given approach, and methods to combine various sources of evidence of pathogenicity. We collaborate with a clinical diagnostic lab to evaluate this evidence to translate it to patients.

Contact: ccassa [at] bwh.harvard.edu (ccassa[at]bwh[dot]harvard[dot]edu)

Location: Longwood (HMS New Research Building)

Current Projects: 

Development of the Human Dynamic Neurochemical Connectome Scanner

The goals of this project are to build, integrate and test the hardware and software components for a next-generation 7 Tesla MR-compatible PET insert with dramatically improved sensitivity and perform proof-of-principle studies in healthy volunteers. 

Total-body PET/CT Imaging

This project aims to implement techniques that take advantage of the latest generation PET/CT scanners, which have an order of magnitude higher sensitivity (e.g., improve quality PET images, reduce radiation exposure and/or imaging acquisition time, etc.), and multi-organ imaging capabilities. 

PET-MR Imaging of pulmonary fibrosis

This project aims to apply PET-MR imaging to quantify molecular abnormalities in the lungs of idiopathic pulmonary fibrosis patients and determine if such measures can predict the pace of disease progression and determine whether the patient is responding to anti-fibrotic therapy.

Contact: ccatana [at] mgh.harvard.edu (ccatana[at]mgh[dot]harvard[dot]edu)

Location: Charlestown

Current Projects: 

Delivery Technologies

Developing protein-based nanoparticles for efficient delivery of therapeutic macromolecules. We are developing protein polymer-based, cell type-specific delivery systems to enhance the efficacy and safety of genome editors, RNA therapeutics, and other macromolecular cargo for in vivo applications, including the treatment of inherited hematopoietic disorders and the design of immuno-oncology therapeutics.

Engineering Living Tissues

Cell and tissue engineering for applications in regenerative medicine. We are investigating genetic pathways that could serve as rational targets to improve the long-term success of organ, tissue, and cell transplantation through multiplex genome editing and developing new additive manufacturing approaches to accelerate organ fabrication. Areas of focus include the development of immunoevasive (hypoimmunogenic) living blood vessels. 

Modulating Innate Immunity

Defining modulators of innate immunity and tissue repair. We are studying transcription factor protein-protein interactions that promote gut tissue integrity and are developing small molecules that target these interactions as immune-modulating therapeutics with relevance to inflammatory bowel disease and other disorders. Areas of focus include regulatory T cell modulators.

Glycobiology & Glycotherapeutics

Defining Clinically Relevant Protein-Glycan Interactions. We are developing tools to study the roles of glycan-protein interactions in innate immunity, thrombosis, and cancer and identifying molecules that target these interactions as therapeutic agents. We hope to decipher underlying mechanisms of cancer-associated venous thromboembolism. 

Biologically Inspired Artificial Organs

Biomolecular engineering to improve the performance of implanted cardiovascular devices and other blood contacting artificial organs. We are developing schemes that regenerate selective bioactive molecular constituents after device implantation to extend the lifetime of anti-thrombogenic films and enhance clinically related performance characteristics of blood contacting devices.

Contact: echaikof [at] bidmc.harvard.edu (echaikof[at]bidmc[dot]harvard[dot]edu)

Location: Longwood

Current Projects: 

Genetics of delayed puberty

We are studying the role of both common and rare genetic variants in causing self-limited delayed puberty (constitutional delay of puberty) to understand the pathways that determine pubertal timing. We are also using genetic instruments to assess the effects of pubertal timing on adult health. 

Genetic testing for differences of sex development (DSD)/intersex conditions

We are analyzing rare-variant causes of DSD's and also studying the impact on parents of receiving genetic testing results. 

Effects of gender-affirming hormonal treatments

As part of the four-site Trans Youth Care (TYC) study, our group is analyzing data on from the largest longitudinal cohort in the US of transgender and nonbinary youth receiving hormonal treatments for gender affirmation (GnRH agonists for pubertal blockade and sex steroids for pubertal induction) to understand physical and psychosocial outcomes of these treatments.

Contact: Yee-Ming.Chan [at] childrens.harvard.edu (Yee-Ming[dot]Chan[at]childrens[dot]harvard[dot]edu)

Location: Longwood 

Current Projects: 

Exploring mitochondrial crista junction biogenesis

The cristae junction (CJ) is a membrane neck that stabilizes the mitochondrial inner-membrane folds, and serves as a nexus of the organelle's diverse functions. How protein assemblies cooperate to mediate this enigmatic membrane structure remain unclear. To answer these questions you will have the opportunity to explore in vitro reconstitution, single particle cryo-EM and/or cryo-ET to understand principles for CJ stabilization and dynamics.

Contact: luke_chao [at] hms.harvard.edu (luke_chao[at]hms[dot]harvard[dot]edu)

Location: MGH Main Campus

Contact: gchurch [at] genetics.med.harvard.edu (gchurch[at]genetics[dot]med[dot]harvard[dot]edu)

Location: Longwood

Current Projects: 

Quantitative hydration status determination

This engineering and clinical project uses MR to provide an absolute measure of hydration status. 

Local microdosing for epilepsy therapy

This project has shown that local dosing can eliminate seizures but avoid the systemic effects of drugs.

Microsampling of interstitial fluid in the brain

This new tool provides a means to obtain proteomic analysis to interstitial fluid in the brain in a longitudinal manner.

Contact: mjcima [at] mit.edu (mjcima[at]mit[dot]edu) | Wendy Brown, webrown [at] mit.edu (webrown[at]mit[dot]edu)

Location: MIT (Bldg. 76-653)

Contact: jimjc [at] mit.edu (jimjc[at]mit[dot]edu)

Location: MIT

Current Projects: 

Multiple projects available relevant to key research areas, please contact for more details.

Contact: alan_dandrea [at] dfci.harvard.edu (alan_dandrea[at]dfci[dot]harvard[dot]edu)

Location: Longwood

Current Projects: 

High-throughput antibody screening technologies and computational approaches for rapid drug discovery

The current low-throughput technologies used for antibody discover cannot deconvolute the molecular features of human immunity against the vast array of potential protein targets. This project will develop and implement new experimental and/or machine learning/AI approaches to comprehensively characterize human antibody immunity, and for improved drug discovery including against global infectious diseases and pandemic health threats. 

Personalized T cell receptor therapies for cancer cures

T cells are known to protect effectively against cancers, but the identification of T cell receptors as drugs for individual patients remains too difficult for clinical use. This project will establish a rapid technology to screen and identify protective T cell receptors and apply it to understand the key features of anti-cancer immune pressure in human patient cohorts and in relevant animal models that recapitulate critical aspects of human disease.

Contact: dekosky [at] mit.edu (dekosky[at]mit[dot]edu)

Location: MIT / The Ragon Institute

Current Projects: 

Projects are designed according to the student's individual interests and skills. Current interests of the lab span a wide range of clinical and biological questions using genomics, transcriptomics, and omics data. In the first phase of the project, students are usually teamed up with a postdoc or senior staff member in the lab to have daily mentorship, in addition to weekly meetings with the PI.

Contact: felix.dietlein [at] childrens.harvard.edu (felix[dot]dietlein[at]childrens[dot]harvard[dot]edu)

Location: Longwood

Current Projects: 

Molecular and structural mechanisms of toxins

Microbiome-pathogen interactions

Develop novel therapeutic proteins for genetic editing and modulation of signaling networks in cells for treating genetic diseases, pain, neurological disorders, and cancer

Contact: min.dong [at] childrens.harvard.edu (min[dot]dong[at]childrens[dot]harvard[dot]edu)

Location: Longwood / Boston Children’s Hospital

Please visit our website and contact us to learn more about our current projects.

Contact: ere [at] mit.edu (ere[at]mit[dot]edu) 

Location: MIT

Current Projects: 

Artificial Intelligence Boosted Evolution and Detection of Genetically Encoded Reporters for In Vivo Imaging

Cell and viral based therapies have the potential to revolutionize the treatment of many diseases. However, the optimization of such biological therapies depends critically on the ability to monitor the spread and persistence of the therapeutic agent and assess the tissue response. This project is focused on developing and optimizing a novel MRI reporter gene technology that allows for the imaging of cell and viral based therapeutics. A novel genetic programing AI has been developed to help predict the optimal reporter protein sequence and structure. The reporter gene technology will be demonstrated for monitoring oncolytic virotherapy in glioblastoma tumor models, however, the technology is generalizable to any cell or viral therapeutic. 

Rapid and Quantitative CEST Imaging with Deep-Learning Magnetic Resonance Fingerprinting

Measurement of tissue pH and protein/metabolite concentrations using chemical exchange saturation transfer (CEST) magnetic resonance imaging (MRI) has been demonstrated to provide crucial insight into many disease pathologies, including tumors, stroke, renal disease, osteoarthritis, and heart failure. However, clinical translation of these CEST-MRI methods has been hindered by the qualitative nature of the image contrast, the long image acquisition times, and the complex data processing required. This project is focused on developing, optimizing and translating to the clinic a novel Magnetic Resonance Fingerprinting (MRF) method that allows for rapid and quantitative pH and protein/metabolite imaging. A machine learning algorithm is being developed for optimizing the MRF acquisition protocol and maximizing the discrimination of different tissue pH and protein/metabolite concentrations.

Contact: cfarrar [at] mgh.harvard.edu (cfarrar[at]mgh[dot]harvard[dot]edu)

Location: Charlestown

Current Projects: 

This collaborative project is to develop a new technology for in-home monitoring of brain waste clearance during sleep. This tool will be used to understand why women have higher risk of Alzheimer's disease, and provide early-onset detection. The project will involve developing a functional near infrared spectroscopy device for home use, validating this tool in the MRI scanner, and applying it in a cohort of women to identify early markers of risk for neurodegenerative disorders.

Contact: mfranceschini [at] mgh.harvard.edu (mfranceschini[at]mgh[dot]harvard[dot]edu) | Laura Lewis, ldlewis [at] mit.edu (ldlewis[at]mit[dot]edu) 

Location: MGH Charlestown & MIT

Please note, this is a collaborative project between PIs

Mission: Our group places a strong emphasis on research which can have an impact on patient care and we train people in the translational process. We emphasize multidisciplinary training and we work closely with clinical collaborators in ophthalmology, surgery, pathology and radiology as well as with international groups and industry. We emphasize developing skills which will enable trainees to become independent researchers and thought leaders. The majority of our previous students and postdocs became faculty in academics and many are upper management in industry. For information on OCT see: https://octnews.org/  

Current Projects: 

Next Generation Optical Coherence Tomograph (OCT) and Methods in Ophthalmology for Detecting and Treating Blinding Diseases

OCT was invented by a former MD, PhD student in our group and has become a standard imaging method in ophthalmology with between 20 and 30 million imaging procedures world wide every year. Our group and clinical collaborators at the New England Eye Center are developing next generation OCT technologies and methods which dramatically extend resolution, speed and functionality. Next generation instruments can acquire gigavoxel class volumetric data, micron scale resolution images, and enable functional as well as structural imaging. Techniques involve optical mechanical design, signal processing, image reconstruction, clinical study design for developing new biomarkers of disease, and studies of disease pathogenesis. We collaborate with internationally recognized clinical thought leaders in ophthalmology, with groups in the US and Europe as well as with industry.

Guiding Cancer Surgery using Nonlinear Microscopy for Real Time Pathology

“The pathologist is the only person who can tell you have you have cancer.” (Quote from Don Coffey, Johns Hopkins) Pathologists detect cancer by using histology, a time consuming technique where tissue specimens are chemically processed, microtomed into thin sections, and stained for examination by transmission light microscopy. Using modern optical techniques including femtosecond lasers, nanometer linear motor stages, high speed optical detection, signal processing, and computation, it is possible to generate histological images of freshly excised surgical specimens in real time. Pathologists can give surgeons information on cancer type and grade as well as surgical margin status to guide decisions during surgery. Real time pathology using nonlinear microscopy has the potential to reduce rates of repeat surgery in breast cancer lumpectomy, reduce rates of incontinence and impotence in prostatectomy, and will have applications to many other cancer surgeries. We collaborate with groups at the Beth Israel Deaconess Medical Center, Harvard Medical School and the Medical University of Vienna.

Contact: jgfuji [at] mit.edu (jgfuji[at]mit[dot]edu) 

Location: MIT

Our laboratory has multiple projects available for dissecting and probing signaling pathways critically important for the function of blood vessels and cardiac tissue. The goal of these efforts is the discovery of novel therapeutics for the treatment of cardiovascular disease. 

Current Projects: 

Identification of novel cellular mechanosensors

This project seeks to identify novel mechanosensors activated in cells in response to flow. It combines a genome-wide CRISPR and small molecule screens with the use in vitro flow systems and in vivo models

Developing new systems to study interactions between blood vessels and cardiac muscle

This project aims to developed cardiac organoids generated with iPSC-derived cells cardiovascular with a perfused vasculature. It involves the use regenerative biology systems combined with experimental platforms including microfluidics and single-cell RNAseq

Contact: guillermo_garcia-cardena [at] hms.harvard.edu (guillermo_garcia-cardena[at]hms[dot]harvard[dot]edu) 

Location: Longwood

Current Projects: 

Deep learning for the microbiome

The microbiome, or trillions of microbes living on and within us, is a complex and dynamic ecosystem that is crucial for human health, and when disrupted may contribute to a variety of diseases including infections, arthritis, allergies, cancer, heart and bowel disorders. The lab develops and applies novel deep learning methods coupled with cutting edge multi-omics approaches (e.g., high-throughput spatial-omics). Applications include resolving microbe-microbe and host-microbe interactions at high spatial resolution, forecasting microbial population dynamics in the gut for rational design of therapies, and predicting the impact of the microbiome on the onset or progression of human diseases.

Contact: ggerber [at] bwh.harvard.edu (ggerber[at]bwh[dot]harvard[dot]edu)

Location: Longwood (Brigham and Women's Hospital)

Please visit our website and contact us to learn more about our current projects. 

Contact: Riaz_Gillani [at] dfci.harvard.edu (Riaz_Gillani[at]dfci[dot]harvard[dot]edu) 

Location: Longwood 

Current Projects: 

Visit the Greka Lab website to learn more about current projects, link below. 

Contact: agreka [at] broadinstitute.org (agreka[at]broadinstitute[dot]org) | Katie Liguori, kliguori [at] broadinstitute.org (kliguori[at]broadinstitute[dot]org)

Location: MIT / The Broad Institute

Current Projects: 

There are several projects available in my lab on the topics of coil design, electromagnetic optimization and neurophysiological modeling. Modeling encompasses both the peripheral nerve and cardiac systems. Applications are MRI gradient coil design and development of novel therapeutical approaches to non-invasive nerve stimulation using magnetic fields. 

Contact: bguerin [at] mgh.harvard.edu (bguerin[at]mgh[dot]harvard[dot]edu)

Location: Charlestown

Current Projects: 

How does an endothelial cell sense shear stress and how does this signaling get altered in disease? 

We have identified key components of the shear stress response pathway, which can cause several vascular diseases. We are studying how co-regulated gene networks change in response to shear and oscillatory blood flow. 

What are the cellular phenotypes that drive genetic risk of cardiovascular disease? 

We use CRISPR screens to identify the pathways regulated by multiple risk loci. First we used Perturb-seq to profile the shared transcriptional effects of CRISPR knockdown of 2000 genes in endothelial cells. Now we are using a technology called Cell Painting to profile the cellular effects of these same CRISPR perturbations. This project would be in close collaboration with the Broad Institute. 

How do the mechanisms of rare and common vascular diseases overlap? 

This project will identify new mechanisms that are shared between coronary artery disease and a rare neurovascular disease, Cerebral Cavernous Malformations. We have identified new mutations that regulate the complex of genes previously implicated in CCM disease. We will sequence patients/families with the disease, and model their mutations in stem cells. We will also test novel ERK5 PROTAC inhibitors developed by our collaborators at Stanford.

Contact: rgupta [at] bwh.harvard.edu (rgupta[at]bwh[dot]harvard[dot]edu) 

Location: Longwood / Broad Institute

Current Projects: 

Imaging the Pulmonary Circulation to Aid Personalized Management of Acute Respiratory Distress Syndrome

In collaboration with the anesthesia team, our research seeks to replicate the condition of human obesity by placing weights on pigs and subsequently inducing acute respiratory distress syndrome (ARDS) through the administration of hydrochloric acid to the lungs. Utilizing advanced portable photon counting CT technology, our team participates in scanning and analyzing the progression and status of ARDS under these conditions, contributing to essential research at the intersection of obesity and respiratory distress. 

The Use of CT Imaging for the Detection and Prognostication of Acute Ischemic Stroke

Our team utilizes the latest software to estimate core and penumbra regions in baseline CTP scans of stroke patients, comparing these initial assessments with final outcomes to evaluate the accuracy and prediction of these markers. 

Enhancing IR Surgical Procedures with Portable PCD Scan Technology 

Collaborating with the IR surgery team, we employ our portable PCD scan to facilitate access to the celiac plexus, electrode placement, and peristalsis stimulation. This project offers significant exposure to cutting-edge IR surgical techniques and demonstrates the integration of advanced imaging technology in complex medical procedures.

Contact: rgupta1 [at] mgh.harvard.edu (rgupta1[at]mgh[dot]harvard[dot]edu)

Location: Charlestown

How do we choose when we can’t know the right answer, or the right answer might not even exist? The philosopher al-Ghazali reasoned free will lies in our ability to act under uncertainty, in situations where we can only pick “randomly” at best. Intriguingly, such (effectively) stochastic behavior may be the key to intelligence. Machine learning algorithms depend on stochasticity, and probabilistic cognitive models explain human reasoning in situations where artificial intelligence struggles. What neural mechanisms enable us to behave so spontaneously? Our lab aims to answer these questions by dissecting the role of neural circuits disrupted in neurological and psychiatric disease. Drawing on clinical insight and intuition, we train mice to do complex behaviors, do neurosurgeries to record and manipulate neural circuits with optogenetics, and use computational modeling to understand how these circuits give rise to spontaneous behavior. 

We have a small lab where you'll work side-by-side with the PI to design and do very cool behavioral neuroscience and computational cognitive science experiments with mice! We're looking for passionate, curious and meticulous HST students eager to learn. Neuroscience experience is great but not required--we will train you! 

We are accepting inquiries from HST students. Unfortunately, we are not reviewing applications from undergraduate researchers, co-ops, technicians or high school students at this time. 

Current Projects: 

Determining the neural circuit mechanisms of spontaneous behavior

We'll train mice to play games and record/manipulate their neurons to understand how animals make spontaneous motor, cognitive and perceptual decisions under uncertainty. 

a. Learn to build electronics and write code to run behavior rigs 

b. Learn to do neurosurgeries in mice 

c. Learn to train mice to play games 

d. Learn to apply electrophysiology, fiber photometry and optogenetics to dissect neural circuits in behaving animals 

e. Learn advanced statistical methods and modeling

Contact: ahamilos [at] wi.mit.edu (ahamilos[at]wi[dot]mit[dot]edu) 

Location: Whitehead Institute

Current Projects: 

ABO Blood Type and the Hemostatic System

ABO blood type is associated with risk for bleeding and thrombosis (clotting). The reason for this epidemiologic association is incompletely understood. We hypothesize that ABO blood group antigens carried on hemostatic glycoproteins have an impact on the function of coagulation factors and platelets. We are working to determine the inventory of hemostatic proteins which carry ABO blood group antigens and how the function of these proteins is affected by ABO blood group status.

Platelet Glycans and the Anti-Platelet Immune Response

Glycans (carbohydrate polymers) on the surface of platelets play an important role in the immune response against platelets. We are working to determine how glycans on the platelet surface affect anti-platelet immune responses in diseases such as immune thrombocytopenia, post-transfusion purpura, and fetal/neonatal alloimmune thrombocytopenia.

Contact: mhollenhorst [at] bwh.harvard.edu (mhollenhorst[at]bwh[dot]harvard[dot]edu) | Letice Arthur, larthur5 [at] bwh.harvard.edu (larthur5[at]bwh[dot]harvard[dot]edu)

Location: Longwood

Current Projects: 

Portable, Low Field Brain Magnetic Resonance Imaging (MRI) for Stroke and Dementia

We are developing machine learning techniques to improve image quality and extract morphometric measures in portable brain MRI, which has huge potential in underserved areas and acute imaging settings (e.g., stroke). 

Deep learning registration & segmentation with NextBrain

We have developed a next-generation histological atlas of the human brain (https://github-pages.ucl.ac.uk/NextBrain) that can be applied to automated segmentation of hundreds of regions in brain MRI scans of living people. However, automated segmentation is slow due to the need of spatially align ("register") the huge atlas to new cases. We are seeking to develop machine learning tools to great speed up this process and make segmentation with NextBrain practical at every site that wants to use it, irrespective of their computational resources. 

Generative models of pathology for clinical brain MRI analysis

Our group has recently developed machine learning techniques for image analysis of brain MRI scans of any resolution and contrast, which has finally enabled the analysis of scans from our hospital PACS. While these techniques work remarkably well, we have so far trained them with relatively normal anatomy. The goal of this project is to learn to synthesize pathology during training (e.g., tumors, strokes) in order to enable the application of our methods in the wild.

Contact: jei [at] mit.edu (jei[at]mit[dot]edu)

Location: Charlestown

Current Projects: 

Digital biomarkers to understand and predict mental states

We have accrued a longitudinal dataset of older adults with daily mood, sleep, and stress ratings, alongside passively collected smartphone sensor data (e.g. accelerometer, geolocation). We are applying machine learning algorithms to understand and predict daily mood on the basis of this data.

Effects of mindfulness and guided imagery on Latino caregiver stress

We are embarking on a new research study with Spanish speaking Latino caregivers, who receive a mobile App that delivers mindfulness and guided imagery alongside caregiver skills. There are opportunities to be involved with the clinical trial implementation as well as with longitudinal data analysis.

Contact: felipe.jain [at] mgh.harvard.edu (felipe[dot]jain[at]mgh[dot]harvard[dot]edu)

Location: MGH

Current Projects: 

Targeting the tumor microenvironment to improve cancer treatment

Our research goals are (i) to understand how the abnormal tumor microenvironment confers resistance to various cancer treatments (e.g., molecular therapeutics, nanotherapeutics, radiation and immunotherapy), (ii) to develop and test new strategies to overcome this resistance, and (iii) to translate these strategies from bench to bedside through multi-disciplinary clinical trials in patients with brain, breast, pancreatic, colon and ovarian cancers. This tight integration between bench and bedside and application of engineering/physical science principles to oncology is a hallmark of our research. 

Contact: rjain [at] mgh.harvard.edu (rjain[at]mgh[dot]harvard[dot]edu) | Elizabeth Garzon, egarzon [at] mgh.harvard.edu (egarzon[at]mgh[dot]harvard[dot]edu)

Location: MGH (Main Campus and Charlestown) 

Current Projects: 

Deciphering the functional role of MKRN3 in puberty and reproduction

We have identified mutations in an imprinted gene, MKRN3, which encodes an E3 ubiquitin ligase, in association with central precocious puberty. Projects are available for students to: (1) Elucidate the potential roles of MKRN3 in neuronal development and synaptic plasticity, using genetically modified mouse models and human iPSC-derived hypothalamic cell models; (2) Examine candidate targets of MKRN3, including KISS1, NKB, IGF2BP1 and LIN28B, as well as novel mRNA targets using eCLIP (enhanced Cross-Linking and ImmunoPrecipitation), in the regulation of the reproductive axis; and (3) Identify new MKRN3 targets and leverage mutations in key protein domains identified in patients with CPP to investigate the roles of different MKRN3 domains in protein function. 

Reproductive biology of gonadotropin regulation

Gonadotropin-releasing hormone (GnRH) is secreted in a pulsatile manner from hypothalamic neurons, and varying GnRH pulse amplitudes and frequencies act on the pituitary to preferentially induce either luteinizing hormone (LH) or follicle-stimulating hormone (FSH) to regulate fertility and the menstrual cycle. Our goal is to identify the mechanisms of this hormone pulse frequency-dependent regulation. Projects are available for students to: (1) Determine the relative contributions of GnRH signaling through specific GnRHR-coupled signaling in gonadotropes to polycystic ovarian syndrome (PCOS), using our previously generated genetically modified mice in a well-established model of PCOS; and (2) Determine the relative contributions of GnRH signaling through specific GnRHR-coupled signaling in gonadotropes to hypothalamic amenorrhea (HA), using our previously generated genetically modified mice in an established model of HA. 

Identifying the functional roles of DLK1 and MeCP2 in puberty and reproduction

We have identified mutations in two additional genes, DLK1 and MeCP2, in association with central precocious puberty. DLK1 (Delta Like Non-Canonical Notch Ligand 1), also known as pre-adipocyte factor 1 (Pref-1) and associated with Temple Syndrome in humans, is a non-canonical ligand in the Notch signaling pathway. MeCP2 (Methyl-CpG-binding protein 2) is an X-linked gene associated with Rett syndrome in humans, encoding a chromatin-associated protein that regulates transcription. Projects are available to determine the mechanisms by which these two genes regulate the timing of puberty, using mouse models as well as in vitro approaches.

Contact: ukaiser [at] bwh.harvard.edu (ukaiser[at]bwh[dot]harvard[dot]edu) | Rona Carroll, rcarroll [at] bwh.harvard.edu (rcarroll[at]bwh[dot]harvard[dot]edu)

Location: Longwood

With our MIT Co-PI, we are developing diverse risk prediction models for Interpersonal violence using historical clinical and imaging data for identifying patients at risk for intimate partner violence, older adult abuse, and non accidental trauma in children. This database offers an excellent opportunity for students to lead smaller, focused research projects (frequently first authored) that contribute to the larger goal of advancing violence prevention and detection. 

Current Projects: 

Injury pattern analysis across various organ systems

Investigating the role of substance use disorders in interpersonal violence 

Examining the intersection of mental health and violence 

Exploring specific diagnoses and their relationship to abuse

Students will have the chance to work with cutting-edge data, collaborate with experts, and contribute meaningful research in a high-impact area of public health. We are building a multimodal Frailty risk prediction model in older adults with our MIT Co-PI using Holistic AI in Medicine. In addition to contributing to the overall development of the ML model, participants will have the opportunity to delve into specific organ imaging components, offering a focused exploration of how imaging data can enhance frailty risk assessment. This project provides a unique blend of cutting-edge AI research and medical imaging analysis, ideal for students interested in the intersection of technology and healthcare.

Contact: bkhurana [at] bwh.harvard.edu (bkhurana[at]bwh[dot]harvard[dot]edu) 

Location: Brigham & Women's Hospital / Remote

Current Projects: 

Risk assessment for breast cancer

Accurate quantification of a woman’s risk of developing invasive breast cancer is critical for appropriate screening of women. To this end, we have access to >300,000 digital breast tomography (DBT) with paired clinical metadata and breast cancer histories. Using this data, we seek to train deep learning-based models to quantify a patient’s 1-year, 2-year, 3-year risk of developing breast cancer. While prior efforts have done this for conventional mammograms, none to date have done this using DBTs. Given our affiliation with MGH, we hope to clinically deploy our models in breast cancer practice. 

Multimodal deep learning to predict therapeutic response

Brain metastases (BM) are an emerging challenge in modern oncology due to increasing incidence and limited treatments. To this end, I recently co-led a phase 2 study that exhibited a 42.1% efficacy rate with pembrolizumab for treatment-refractory BM of diverse tumor types. Given the high toxicity rate of ICI, there is a critical unmet need to develop improved biomarkers for ICI response to minimize harm for patients who are unlikely to benefit from ICI. Therefore, we seek to design deep learning models (DL) that analyze and integrate MRI, histopathology (H&E), and clinico-genomic data from a biobank of 2100 BM patients to create next generation biomarkers that identify BM patients that benefit from ICI. The premise for multimodal data integration is that different modalities provide complementary biological data and together have predictive value beyond that of any individual modality. 

Deep learning-based classification using histopathology whole slide images

Our group has data demonstrating that deep learning can estimate and quantify different facets of the tumor microenvironment – which is critical for determining therapeutic efficacy. To this end, we have datasets in brain, breast, and prostate cancer from both Boston and lower-middle income countries (Nigeria, Haiti). We have many available projects in training deep learning models, using histopathology, to determine clinically impactful biological endpoints.

Location: Charlestown

Contact: akim46 [at] mgh.harvard.edu (akim46[at]mgh[dot]harvard[dot]edu) | cbridge [at] mgh.harvard.edu (cbridge[at]mgh[dot]harvard[dot]edu) 

Our lab’s mission is to gain molecular insight into the differential regenerative ability of mammalian organs so that we can modulate the regenerative capacity of any organ in the setting of disease. To make this possible, our lab is pursuing two parallel but complementary agendas. First, we are innovating tools for high-throughput functional genomics in mouse tissues to bring these longstanding questions of regeneration within reach. Second, we are leveraging these technologies alongside other approaches to uncover the molecular rules governing permissive and restrictive contexts for regeneration.  

Current Projects: 

We believe in working with each trainee to develop their own unique project that suits their interests, skills, and goals. 

Bringing high-throughput functional genomics into the organism

We recently established genome-wide CRISPR screening in the mouse liver (Keys and Knouse 2022). Although genome-wide screening in the liver alone can offer novel insights into diverse phenomena, the full potential of high-throughput screening in the organism rests on expanding this technology to other organs and CRISPR applications. We are thus keen to establish efficient and stable transgene delivery and genome-scale CRISPR screening in other cell types and organs of interest in the living mouse. 

Identifying the molecular requirements for organ regeneration

The liver is the only solid organ with the capacity to regenerate itself. However, the molecular rules governing this uniquely permissive context for regeneration are unknown. We are keen to apply our in vivo genome-wide CRISPR screening platforms to identify the genes that endow the liver with its remarkable regenerative capacity as well as the genes that prohibit regeneration in other organs. By identifying these genes, we will bring our goal of conferring regenerative capacity to other tissues into the realm of possibility.

Contact: knouse [at] mit.edu (knouse[at]mit[dot]edu)

Location: MIT

Current Projects: 

Real-time Fair Machine Learning for Clinical Decision Support to Reduce Health Disparities 

We aim to develop a class of machine learning prediction algorithms that can adapt to changing hospital environments in real time and make accurate predictions among patient subpopulations. Possible applications include patient risk predictions in emergency rooms, and admission predictions during epidemics. Students should have familiarity with Python and an interest in machine learning for health. 

Exploring the Fairness-Accuracy Trade-off in Machine Learning Models 

We aim to draw recent sampling methods from statistics to generate a collection of machine learning models which are fair as well as nearly optimal. Seeking students familiar with Python or Julia to help us code some sampling methods and help test novel algorithms. 

Foundation Models for Heart Health

We aim to build foundation models out of a large set of electrocardiograms and echocardiograms to better understand and predict various heart-related outcomes in pediatric populations. Seeking students passionate about machine learning in Python and interested in cutting edge deep learning approaches that learn from multi-modal data.

Contact: william.lacava [at] childrens.harvard.edu (william[dot]lacava[at]childrens[dot]harvard[dot]edu) 

Location: Fenway

Current Projects: 

Brain Organoid MAP (Microphysiological Analysis Platforms) 

EXODUS (EXOsome Detection via the Ultrafast-purification System) 

Brain-Muscle Microphysiological System for Understanding Aging in Space

QBET (Quantum Biological Electron Transfer) Imaging in living cells

Contact: lplee [at] bwh.harvard.edu (lplee[at]bwh[dot]harvard[dot]edu) 

Location: Longwood

Current Projects: 

Analyzing fast, multimodal neuroimaging data

Noninvasive tools such as fMRI and EEG allow us to measure activity in the human brain, but classically they have limited spatial and temporal resolution. We develop signal processing and computational approaches to discover new information in human neuroimaging data. For example, we use signal processing strategies to identify fast information in high temporal resolution fMRI data and develop machine learning and analysis strategies for integrating EEG and fMRI dynamics.

Fluid dynamics of the human brain

We develop new MRI methods to measure cerebrospinal fluid flow and vascular physiology in the human brain, to understand the flow of cerebrospinal fluid and its relevance for brain waste transport. 

Identifying neural circuits that control sleep

We are using novel neuroimaging techniques to identify the deep-brain neural circuits that control sleep in humans. We develop analysis methods to allow us to measure new aspects of neural circuits with ultra-high field fMRI and apply these to identify brain systems regulating sleep. 

Sleep and brain- and body-wide physiology

Sleep transforms the basic physiological processes of the brain and body. We are using multimodal neuroimaging to understand how sleep drives waves of fluid flow over the brain, and how sleep modulates body-wide physiology, such as changes in blood vessels, blood pressure, breathing, and inflammation. 

The sleeping brain in healthy aging and in Alzheimer’s disease

Healthy sleep enhances cognition, and sleep disruption is linked to serious neurological disorders such as Alzheimer’s disease. We are imaging individuals at risk for Alzheimer’s disease to understand how sleep contributes to healthy aging. 

Computational neuroscience of brainwide dynamics underlying natural sleep behaviors

The sleeping brain is highly active: it consolidates information and enhances learning of tasks performed during wakefulness. In natural environments (rather than structured artificial tasks), behavior is complex, and we aim to understand how sleep supports learning of complex behaviors. This project involves machine learning approaches for analysis of latent dynamics in extremely high-dimensional longitudinal electrophysiological recordings from freely behaving and sleeping mice.

Contact: ldlewis [at] mit.edu (ldlewis[at]mit[dot]edu)

Location: MIT

Please visit our website and contact us to learn more about our current projects.

Contact: tami [at] mit.edu (tami[at]mit[dot]edu) 

Location: MIT

Current Projects: 

Spatial immunology tools

Develop and apply advanced technologies to profile T-cell receptors (TCRs), B-cell receptors (BCRs), and viral transcripts within tissue contexts. Use spatial transcriptomics to map immune responses (in particular germinal center responses) in human tissues and mouse models, uncovering and validating new immune mechanisms and therapeutic targets. 

Human immune aging in tissues - understanding, regenerating, and reprogramming

Study the cellular and molecular mechanisms of immune aging using human bone marrow, thymus, and lymph node samples. Explore innovative approaches to regenerate and reprogram aged immune tissues, aiming to enhance immune function and resilience in the aging population. 

Immunological isoform discovery

Use cutting-edge long-read sequencing technologies to discover and characterize novel immunological isoforms in various immune tissues. Identify new biomarkers and therapeutic targets, contributing to a deeper understanding of immune regulation, memory, and diversity. 

Host-pathogen infection tracing

Develop new mouse models for labeling cell interactions and tracing host-pathogen infections in vivo and in tissues. This project aims to provide detailed insights into the dynamics of host-pathogen interactions and inform strategies for combating infections.

Contact: sophia.liu [at] mgh.harvard.edu (sophia[dot]liu[at]mgh[dot]harvard[dot]edu)

Location: MIT / The Ragon Institute 

Current Projects: 

Predictive/prognostic AI in oncology

We have several projects focusing on predicting patient prognosis and treatment response from multimodal data using AI, with an end goal of helping optimize treatment strategies for each patient.

Science of clinical AI translation

We are broadly interested in assessing and improving how AI is translated to clinical practice, including recent efforts in explainability, generalization, bias, and clinician-AI interaction. 

Contact: lotterb [at] ds.dfci.harvard.edu (lotterb[at]ds[dot]dfci[dot]harvard[dot]edu)

Location: Longwood