6.7930_[6.871]/HST.956: Machine Learning for Healthcare

Instructor: Peter Szolovits
Teaching Assistants: Helen Bang, Shibal Ibrahim, Subha Nawer Pushpita
Graduate level; Units 4-0-8 (counts as an AUS2, AAGS, and II subject; also a EECS AI TQE)
Lectures: Tuesdays & Thursdays, 2:30-4:00pm Eastern Time, 54-100

Finally, as of Feb 27, 54-100 is ready for use, so going forward, we will meet in that room.
One more disruption: On March 21, 54-100 is in use for a dedication event, so we will meet that day (only) in 10-250.

Recitations (required): Friday, 3:00-4:00pm, 4-270
Prerequisite: 6.3900_[6.036] or 6.7900_[6.867] or 9.520/6.7910_[6.860] or 6.8611/6.8610_{[6.806/6.864]} or 6.4102/6.4100_{[6.438/6.034]} or equivalent machine learning class. (Subscripted bracketed numbers are the class numbers before the recent mass renumbering of all EECS classes.)
Office Hours, most weeks:

When	Where	Who
Mon 4-5	36-112	Shibal
Tue 1:30-2:30	38-166	Pushpita
Wed 4:30-5:30	34-301	Helen

Contact staff: mlhc2024@googlegroups.com

Announcements

Lectures, lecture slides, readings, recitation material, and assignments will be posted to our Canvas site
We expect you to attend all lectures and recitations, and will do a virtual roll call for each using the following links:

When readings are posted for a lecture, we expect you to provide us a brief summary of that reading by the day after the corresponding lecture. Three relevant bullet points are sufficient. Please submit these at the following link:

Reading summary submission

Please accept your email invitation or add yourself to the course Piazza (find the link in a Canvas announcement)
We do not plan to support remote participation in lectures and our expectation is that students will attend class in person unless an exception is granted (e.g., students needing to quarantine/isolate, or a clinician who is on service).
See last year's site for the 2023 schedule and problem sets; this year's will be similar, but not identical. See that site for slides or MIT OpenCourseWare for videos of past lectures.
This class will have a final exam, but no midterm.
Class group projects will be organized with clinician mentors. Students will present their results in a poster session during the last week of class.

Course description

Introduces students to machine learning in healthcare, including the nature of clinical data and the use of machine learning for risk stratification, disease progression modeling, precision medicine, diagnosis, subtype discovery, and improving clinical workflows. Topics include causality, interpretability, algorithmic fairness, time-series analysis, graphical models, deep learning and transfer learning. Guest lectures by clinicians from the Boston area and course projects with real clinical data emphasize subtleties of working with clinical data and translating machine learning into clinical practice.

Schedule

The schedule for classes is under significant revision throughout the semester.

Please visit our Canvas site to access all slide decks, recitation materials, and more.

	Class	Date	Lecture & Materials	Assignments
Overview of Clinical Care & Data	1	Tue, Feb 6	Introduction: What makes healthcare unique? Reading: Machine Learning in Medicine
	2	Thu, Feb 8	Overview of Clinical Care Reading: AI in health and medicine	PS 0 out
	3	Tue, Feb 13	Overview of Clinical Data Science Reading: EHR Safari Data is Contextual	PS0 due
ML with Clinical Text, Imaging, Physiological, and Administrative Data	4	Thu, Feb 15	Risk Stratification from Structured Health Data Reading: Razavian N, Blecker S, Schmidt AM, Smith-McLallen A, Nigam S, Sontag D (2015) Population-level prediction of type 2 diabetes from claims data and analysis of risk factors. Big Data 3:4, 277-287, DOI: 10.1089/big.2015.0020.	ps1 out
		Tue, Feb 20	No class -- Monday schedule of classes
	5	Thu, Feb 22	Cautionary Tales; Discussion of Project Ideas; Guest speaker Dr. Leo Anthony Celi, BIDMC & IMES-LCP Description of possible class projects; Dr. Celi and Dr. Alexander Dale, MIT Solve Director of Global Programs. Reading: Joining the battle against health care bias. MIT News, May 16, 2023
	6	Tue, Feb 27	Risk Stratification as Regression; and Physiological Time-Series Reading: Constrained transformer network for ECG signal processing and arrhythmia classification	see potential project list
	7	Thu, Feb 29	Intro to Clinical NLP Reading: Beam AL, Kompa B, Schmaltz A, Fried I, Weber G, Palmer N, et al. Clinical Concept Embeddings Learned from Massive Sources of Multimodal Medical Data. In: Biocomputing 2020, p. 295-306.	ps1 due ps2 out
	8	Tue, Mar 5	Contemporary Clinical NLP Methods [Pushpita] Reading: Huang K, Altosaar J, Ranganath R. ClinicalBERT: Modeling Clinical Notes and Predicting Hospital Readmission [Internet]. arXiv; 2020. Helpful optional readings: The Illustrated GPT-2 (Visualizing Transformer Language Models) A Visual Guide to Using BERT for the First Time	fill our project preference form
	9	Thu, Mar 7	Survival Analysis, Censoring, Proportional Hazard Models Reading: Deep Cox Mixtures for Survival Regression
Causal Inference	10	Tue, Mar 12	Causal Inference, Conditional Treatment Effects Reading for lectures 10 and 11 (feedback form to be submitted after lecture 11): Causal Inference in the Health Sciences: A Conceptual Introduction
	11	Thu, Mar 14	Causal Inference, continued, and Intro to Reinforcement Learning	ps2 due ps3 out
	12	Tue, Mar 19	Causal Inference, continued; policies, instrumental variables, bias from unmeasured confounders No additional reading
	13	Thu, Mar 21	Dataset Shift Reading: Ji CX, Alaa AM, Sontag D. Large-Scale Study of Temporal Shift in Health Insurance Claims. In: Proceedings of Machine Learning Research, 2023. p. 243–78. (feedback can be submitted until Sunday, 11:59pm) (Class meets in 10-250!)
Real World Deployment Challenges		Mar 25-29	Spring Break -- No classes
	14	Tue, Apr 2	Learning with Noisy Labels, Weak Supervision & In-context learning [Pushpita] Reading: Halpern Y, Choi Y, Horng S, Sontag D. Using anchors to estimate clinical state without labeled data. AMIA Annual Symposium proceedings. 2014;2014:606–15.	ps3 due
	15	Thu, Apr 4	The future of EHRs; guest lecture by Dr. Monica Agrawal Reading: Gawande, A. Why Doctors Hate Their Computers, New Yorker, Nov 5, 2018.	ps4 out
	16	Tue, Apr 9	ML for Drug Discovery and Repurposing; Guest lecture by Prof. Jim Collins Reading: Stokes JM, Yang K, Swanson K, Jin W, Cubillos-Ruiz A, Donghia NM, et al. A Deep Learning Approach to Antibiotic Discovery. Cell. 2020 Feb;180(4):688-702.e13.
	17	Thu, Apr 11	Human-AI Collaboration in Clinical ML; guest lecture by Hussein Mozannar Reading: Human-computer collaboration for skin cancer recognition
	18	Tue, Apr 16	Interpretability [Shibal and Helen] Reading: Molnar, C. Interpretable Machine Learning: A Guide for Making Black Box Models Explainable. Read Chapters 5 and 9.	ps4 due
	19	Thu, Apr 18	ML for Medical Imaging: algorithms & applications; guest lecture by Dr. Alexander Goehler Readings: Garcia-Layana A, Cabrera-López F, García-Arumí J, Arias-Barquet L, Ruiz-Moreno JM. Early and intermediate age-related macular degeneration: update and clinical review. CIA. 2017 Oct;Volume 12:1579–87. [Optional] Kappos L, Wolinsky JS, Giovannoni G, Arnold DL, Wang Q, Bernasconi C, et al. Contribution of Relapse-Independent Progression vs Relapse-Associated Worsening to Overall Confirmed Disability Accumulation in Typical Relapsing Multiple Sclerosis in a Pooled Analysis of 2 Randomized Clinical Trials. JAMA Neurol. 2020 Sep 1;77(9):1132.
	20	Tue, Apr 23	Genomics in Medicine: practices and ethics; guest lecture by Sarah Faye Gurev, Courtney Shearer, and Rose Orenbuch Reading: Spiekerkoetter U, Bick D, Scott R, Hopkins H, Krones T, Gross ES, et al. Genomic newborn screening: Are we entering a new era of screening? J of Inher Metab Disea. 2023 Sep;46(5):778–95.
	21	Thu, Apr 25	Methods for genomics in medicine; guest lecture by Sarah Faye Gurev, Courtney Shearer, and Rose Orenbuch Reading: Chen Y, Paramo MI, Zhang Y, Yao L, Shah SR, Jin Y, et al. Finding Needles in the Haystack: Strategies for Uncovering Noncoding Regulatory Variants. Annu Rev Genet. 2023 Nov 27;57(1):201–22.
	22	Tue, Apr 30	Fairness; Guest lecture by Prof. Marzyeh Ghassemi Reading: Chen IY, Pierson E, Rose S, Joshi S, Ferryman K, Ghassemi M. Ethical Machine Learning in Healthcare. Annu Rev Biomed Data Sci. 2021 Jul 20;4(1):123–44.
	23	Thu, May 2	Flipping the clinic with AI: Medicine in the absence of primary care; Guest lecture by Dr. Isaac Kohane, Harvard Dept. of Biomedical Informatics Rosenbaum L. On Calling — From Privileged Professionals to Cogs of Capitalism? N Engl J Med. 2024 Feb;390(5):471–5.
	24	Tue, May 7	Privacy and Confidentiality Reading: The Belmont Report (You probably read this for your CITI training.)
	25	Thu, May 9	Regulation, Law and Deployment; Guest lecture by Christopher Conley, JD, BU Law School Readings: Good Machine Learning Practice for Medical Device Development: Guiding Principles (short overview) Artificial Intelligence & Medical Products
	26	Tue, May 14	Student Project Presentations -- Grier Room (34-401), using posters
		May 20, 9am-noon	Final Exam at Johnson Ice Rink

Reading Assignments

Many of the lectures are associated with related papers that should help you think about the lecture topic. For each reading assignment, you are expected to submit a brief summary of the three most important ideas of the paper, as short bullet points.

Problem sets

The problem sets pdfs will be available here (not that some data for the problem sets is not publicly available):

Some of the recitation materials will be available here:

Projects

We are recruiting a group of doctors with interesting clinical problems to mentor teams of students who will work on them. We will form teams and match to problems/mentors a few weeks into the class. We will release the project guidelines and detailed information in Canvas. For now, please look at the list of proposed projects here.

Grading

40% course project
30% homework (~4 problem sets; both theory & practice)
20% final exam
5% participation – note: class attendance is required (with obvious exceptions for illness, etc.)
5% reading responses

Late Policy (starting for pset1 onwards)

[4 "slack" days] We understand that sometimes things outside one's control prevent submitting by the deadline. As such, each student is given 4 "slack" days that they can use throughout the semester (e.g. you could submit four psets one day late each or you could submit one pset three days late and another one day late) without a late penalty. The days do not subdivide into sub-day units: 2 hours late would spend one of the slack days without 22 hours of "rollover". In your pdf writeup, specify how many slack days you are using (they cannot be used retroactively).
[10% off per unexcused late day.] If you submit a pset 3 days late and use 1 slack day, then this is 2 unexcused late days, which translates to 20% off your homework.
[Max 4 days late.] Homework will not be accepted after 4 days late, regardless of how many slack days are used, absent communication from S3 or OGE.
[write on homework] In order to use a slack day, students must include it in writing on their submission pdf. Otherwise, TAs will assume no slack days used and deduct 10% for each late day.

Scenarios:

Sam uses 2 slack days on HW3. This is the first time Sam has used any slack days. Sam now has 2 remaining slack days and receives her homework score with no penalty.
Jamie uses 1 slack day on HW3 but submits 52 hours after the deadline. Therefore Jamie is 3 days late (rounded up) and receives 20% off the graded homework. This is the first time Jamie has used any slack days, so Jamie now has 1 slack day remaining.
Cory has never used any slack days, but submits HW5 100 hours late with no note / communication from S3 or OGE. This is more than 4 days after the deadline, so the work is not accepted, and Cory receives a zero.

Use of Generative AI

Although you may choose to use generative AI tools to help think through problems, any work you turn in is your responsibility. Pedagogically, you will also learn much more from working through any problem than from copying its answer from a tool or other source. We may ask you to explain the rationale for an answer if it appears that it is not original to you.

6.7930[6.871]/HST.956: Machine Learning for Healthcare