About

Peter Szolovits is a Professor of Computer Science and Engineering at MIT, specializing in Artificial Intelligence and Decision-making. His research areas include AI for Healthcare and Life Sciences, Artificial Intelligence and Machine Learning, and Biological and Medical Devices and Systems. He combines expertise from computer science and electrical engineering to develop systems that interact with the external world through perception, communication, and action, focusing on learning, decision-making, and adaptation in changing environments. Szolovits's work leverages computational, theoretical, and experimental tools to address shared human challenges, particularly in healthcare applications.

Research topics

• Computer science
• Artificial intelligence
• Medicine
• Machine learning
• Natural language processing

Selected publications

• An open natural language processing (NLP) framework for EHR-based clinical research: a case demonstration using the National COVID Cohort Collaborative (N3C)

  UNC Libraries · 2025-03-18

  articleOpen access

  Despite recent methodology advancements in clinical natural language processing (NLP), the adoption of clinical NLP models within the translational research community remains hindered by process heterogeneity and human factor variations. Concurrently, these factors also dramatically increase the difficulty in developing NLP models in multi-site settings, which is necessary for algorithm robustness and generalizability. Here, we reported on our experience developing an NLP solution for Coronavirus Disease 2019 (COVID-19) signs and symptom extraction in an open NLP framework from a subset of sites participating in the National COVID Cohort (N3C). We then empirically highlight the benefits of multi-site data for both symbolic and statistical methods, as well as highlight the need for federated annotation and evaluation to resolve several pitfalls encountered in the course of these efforts.

  PublisherDOI

• Toward responsible AI governance: Balancing multi-stakeholder perspectives on AI in healthcare

  International Journal of Medical Informatics · 2025-06-19 · 14 citations

  article
  PublisherDOI

• Reflections from Research Roundtables at the Conference on Health, Inference, and Learning (CHIL) 2025

  ArXiv.org · 2025-10-17

  preprintOpen access

  The 6th Annual Conference on Health, Inference, and Learning (CHIL 2025), hosted by the Association for Health Learning and Inference (AHLI), was held in person on June 25-27, 2025, at the University of California, Berkeley, in Berkeley, California, USA. As part of this year's program, we hosted Research Roundtables to catalyze collaborative, small-group dialogue around critical, timely topics at the intersection of machine learning and healthcare. Each roundtable was moderated by a team of senior and junior chairs who fostered open exchange, intellectual curiosity, and inclusive engagement. The sessions emphasized rigorous discussion of key challenges, exploration of emerging opportunities, and collective ideation toward actionable directions in the field. In total, eight roundtables were held by 19 roundtable chairs on topics of "Explainability, Interpretability, and Transparency," "Uncertainty, Bias, and Fairness," "Causality," "Domain Adaptation," "Foundation Models," "Learning from Small Medical Data," "Multimodal Methods," and "Scalable, Translational Healthcare Solutions."

  PublisherOA PDFDOI

• Large Language Models Seem Miraculous, but Science Abhors Miracles

  NEJM AI · 2024-05-09 · 9 citations

  article1st authorCorresponding

  Generative artificial intelligence models exhibit amazing abilities but make serious errors. We have a very limited understanding of why they work well at all or of the circumstances under which they give incorrect responses. This suggests the need for additional research and great caution in deploying such models for critical applications. Since the availability of ChatGPT in late 2022, based on OpenAI's GPT 3.5 large language model, those of us who have explored its capabilities have been amazed by its facility with language and its abilities to generate coherent — and even insightful — synopses; answer questions about everything from general knowledge to domain-specific topics; offer advice on how to accomplish tasks, including for medical diagnosis, therapy, and prognosis; deduce consequences of assumptions; and even write effective computer programs. Nevertheless, I would urge great caution in adopting such methods in health care, mainly because of our lack of understanding of how they accomplish the miraculous-seeming things they are able to do.

  PublisherDOI

• An open natural language processing (NLP) framework for EHR-based clinical research: a case demonstration using the National COVID Cohort Collaborative (N3C)

  Journal of the American Medical Informatics Association · 2023-08-08 · 24 citations

  articleOpen access

  Despite recent methodology advancements in clinical natural language processing (NLP), the adoption of clinical NLP models within the translational research community remains hindered by process heterogeneity and human factor variations. Concurrently, these factors also dramatically increase the difficulty in developing NLP models in multi-site settings, which is necessary for algorithm robustness and generalizability. Here, we reported on our experience developing an NLP solution for Coronavirus Disease 2019 (COVID-19) signs and symptom extraction in an open NLP framework from a subset of sites participating in the National COVID Cohort (N3C). We then empirically highlight the benefits of multi-site data for both symbolic and statistical methods, as well as highlight the need for federated annotation and evaluation to resolve several pitfalls encountered in the course of these efforts.

  PublisherOA PDFDOI

• Using Machine Learning to Develop Smart Reflex Testing Protocols

  arXiv (Cornell University) · 2023-02-01 · 2 citations

  preprintOpen access

  Objective: Reflex testing protocols allow clinical laboratories to perform second line diagnostic tests on existing specimens based on the results of initially ordered tests. Reflex testing can support optimal clinical laboratory test ordering and diagnosis. In current clinical practice, reflex testing typically relies on simple "if-then" rules; however, this limits their scope since most test ordering decisions involve more complexity than a simple rule will allow. Here, using the analyte ferritin as an example, we propose an alternative machine learning-based approach to "smart" reflex testing with a wider scope and greater impact than traditional rule-based approaches. Methods: Using patient data, we developed a machine learning model to predict whether a patient getting CBC testing will also have ferritin testing ordered, consider applications of this model to "smart" reflex testing, and evaluate the model by comparing its performance to possible rule-based approaches. Results: Our underlying machine learning models performed moderately well in predicting ferritin test ordering and demonstrated greater suitability to reflex testing than rule-based approaches. Using chart review, we demonstrate that our model may improve ferritin test ordering. Finally, as a secondary goal, we demonstrate that ferritin test results are missing not at random (MNAR), a finding with implications for unbiased imputation of missing test results. Conclusions: Machine learning may provide a foundation for new types of reflex testing with enhanced benefits for clinical diagnosis and laboratory utilization management.

  PublisherOA PDFDOI

• Right, No Matter Why: AI Fact-checking and AI Authority in Health-related Inquiry Settings

  arXiv (Cornell University) · 2023-10-22 · 1 citations

  preprintOpen accessSenior author

  Previous research on expert advice-taking shows that humans exhibit two contradictory behaviors: on the one hand, people tend to overvalue their own opinions undervaluing the expert opinion, and on the other, people often defer to other people's advice even if the advice itself is rather obviously wrong. In our study, we conduct an exploratory evaluation of users' AI-advice accepting behavior when evaluating the truthfulness of a health-related statement in different "advice quality" settings. We find that even feedback that is confined to just stating that "the AI thinks that the statement is false/true" results in more than half of people moving their statement veracity assessment towards the AI suggestion. The different types of advice given influence the acceptance rates, but the sheer effect of getting a suggestion is often bigger than the suggestion-type effect.

  PublisherOA PDFDOI

• Coding Inequity: Assessing GPT-4’s Potential for Perpetuating Racial and Gender Biases in Healthcare

  medRxiv · 2023-07-16 · 30 citations

  preprintOpen access

  Abstract Background Large language models (LLMs) such as GPT-4 hold great promise as transformative tools in healthcare, ranging from automating administrative tasks to augmenting clinical decision- making. However, these models also pose a serious danger of perpetuating biases and delivering incorrect medical diagnoses, which can have a direct, harmful impact on medical care. Methods Using the Azure OpenAI API, we tested whether GPT-4 encodes racial and gender biases and examined the impact of such biases on four potential applications of LLMs in the clinical domain—namely, medical education, diagnostic reasoning, plan generation, and patient assessment. We conducted experiments with prompts designed to resemble typical use of GPT-4 within clinical and medical education applications. We used clinical vignettes from NEJM Healer and from published research on implicit bias in healthcare. GPT-4 estimates of the demographic distribution of medical conditions were compared to true U.S. prevalence estimates. Differential diagnosis and treatment planning were evaluated across demographic groups using standard statistical tests for significance between groups. Findings We find that GPT-4 does not appropriately model the demographic diversity of medical conditions, consistently producing clinical vignettes that stereotype demographic presentations. The differential diagnoses created by GPT-4 for standardized clinical vignettes were more likely to include diagnoses that stereotype certain races, ethnicities, and gender identities. Assessment and plans created by the model showed significant association between demographic attributes and recommendations for more expensive procedures as well as differences in patient perception. Interpretation Our findings highlight the urgent need for comprehensive and transparent bias assessments of LLM tools like GPT-4 for every intended use case before they are integrated into clinical care. We discuss the potential sources of these biases and potential mitigation strategies prior to clinical implementation.

  PublisherOA PDFDOI

• Structure-inducing pre-training

  Nature Machine Intelligence · 2023-06-01 · 20 citations

  articleOpen access

  Abstract Language model pre-training and the derived general-purpose methods have reshaped machine learning research. However, there remains considerable uncertainty regarding why pre-training improves the performance of downstream tasks. This challenge is pronounced when using language model pre-training in domains outside of natural language. Here we investigate this problem by analysing how pre-training methods impose relational structure in induced per-sample latent spaces—that is, what constraints do pre-training methods impose on the distance or geometry between the pre-trained embeddings of samples. A comprehensive review of pre-training methods reveals that this question remains open, despite theoretical analyses showing the importance of understanding this form of induced structure. Based on this review, we introduce a pre-training framework that enables a granular and comprehensive understanding of how relational structure can be induced. We present a theoretical analysis of the framework from the first principles and establish a connection between the relational inductive bias of pre-training and fine-tuning performance. Empirical studies spanning three data modalities and ten fine-tuning tasks confirm theoretical analyses, inform the design of novel pre-training methods and establish consistent improvements over a compelling suite of methods.

  PublisherOA PDFDOI

• Do We Still Need Clinical Language Models?

  arXiv (Cornell University) · 2023-02-16 · 50 citations

  preprintOpen access

  Although recent advances in scaling large language models (LLMs) have resulted in improvements on many NLP tasks, it remains unclear whether these models trained primarily with general web text are the right tool in highly specialized, safety critical domains such as clinical text. Recent results have suggested that LLMs encode a surprising amount of medical knowledge. This raises an important question regarding the utility of smaller domain-specific language models. With the success of general-domain LLMs, is there still a need for specialized clinical models? To investigate this question, we conduct an extensive empirical analysis of 12 language models, ranging from 220M to 175B parameters, measuring their performance on 3 different clinical tasks that test their ability to parse and reason over electronic health records. As part of our experiments, we train T5-Base and T5-Large models from scratch on clinical notes from MIMIC III and IV to directly investigate the efficiency of clinical tokens. We show that relatively small specialized clinical models substantially outperform all in-context learning approaches, even when finetuned on limited annotated data. Further, we find that pretraining on clinical tokens allows for smaller, more parameter-efficient models that either match or outperform much larger language models trained on general text. We release the code and the models used under the PhysioNet Credentialed Health Data license and data use agreement.

  PublisherOA PDFDOI

Recent grants

• NIH Grant R24RR001320

  NIH · $146k · 1987

• NIH Grant R01LM004493

  NIH · $6.2M · 1994

• NIH Grant R01LM005296

  NIH · $697k · 1997

• NIH Grant R01LM009723

  NIH · $1.4M · 2012

• NIH Grant R01HL033041

  NIH · $3.5M · 1998

Frequent coauthors

• Isaac S. Kohane

  Harvard University

  128 shared

• Shawn N. Murphy

  83 shared

• Katherine P. Liao

  Harvard University

  76 shared

• Tianxi Cai

  Harvard University

  75 shared

• Stanley Y. Shaw

  49 shared

• Susanne Churchill

  Harvard University

  48 shared

• Ashwin N. Ananthakrishnan

  Harvard University

  48 shared

• Vivian S. Gainer

  48 shared

Labs

• MIT EECS Artificial Intelligence + Decision-making LabPI

Education

• PhD, Information Science

  California Institute of Technology

  1974

• BS, Physics

  California Institute of Technology

  1970

Awards & honors

• 2021 EECS Awards