Biostatistics · Yale University

Harsh Parikh

Assistant Professor of Biostatistics

Yale School of Public Health

I develop causal inference methods to study how interventions affect people, and how evidence transfers across populations and settings.

Explore my research View CV
Email Google Scholar GitHub LinkedIn
Harsh Parikh
CEADSCausal Evidence & Decisions Studio

The latest

Recent news

New papers, acceptances,
and talks.

  1. Workshop talk

    Lightning talk at the University of Michigan's Causal AI for Decision Making Workshop.

  2. Journal acceptance

    Demystifying Prediction-Powered Inference was accepted by the Harvard Data Science Review.

  3. New preprint

    Proximal causal inference through cross-proxy balancing is available on arXiv.

  4. Journal acceptance

    Demystifying Prediction Powered Inference received provisional acceptance at the Harvard Data Science Review.

  5. NeurIPS 2026

    Controllable Generative Sandbox for Causal Inference was accepted as a poster.

  6. NeurIPS 2026

    Characterizing Underrepresentation in Generalizing Causal Survival Estimates was accepted as a poster.

  7. Published

    The epidemiology of artificial intelligence was published in Nature Health.

  8. Journal acceptance

    Data (in)equities in data science was accepted by the Canadian Journal of Statistics.

About

Causal evidence for decisions.

I lead the Causal Evidence and Decisions Studio (CEADS) at Yale. We combine statistics and machine learning to estimate the effects of interventions, with particular attention to differences between study samples and the populations those interventions are intended to serve.

Our work connects methodological questions to problems in public health, medicine, economics, and ecology. Applications include the economic costs of illness, treatment decisions in critical care, and the effects of agriculture on forests in Madagascar.

I received my PhD in Computer Science from Duke University and completed a postdoctoral fellowship in Biostatistics at Johns Hopkins.

Full background and affiliations

Research

Methods, with applications.

Estimating causal effects when data are incomplete, populations differ, and decisions have consequences.

Data fusion & generalizability

Combining experimental and observational studies, assessing underrepresentation, and estimating effects across populations, time, and social networks.

Who are we missing? ↗Data fusion ↗

Interpretable causal learning

Developing matching methods, treatment rules, and approaches to extrapolation that make the relationship between data and causal estimates easier to examine.

MALTS ↗Regularizing extrapolation ↗

Publications

Papers & preprints

Full publication list in my CV

2026

Nature Health

The Epidemiology of Artificial Intelligence

H Parikh, T McCormick, EK Johnson, L Hickey, ML Ranney, B Mukherjee
Observational Studies

A Double Machine Learning Approach for Combining Experimental and Observational Studies

H Parikh, M Morucci, V Orlandi, S Roy, C Rudin, A Volfovsky
JNCI Cancer Spectrum

Frailty and Low Utilization of Curative-Intent Surgical Resection in Non-Metastatic Pancreatic Cancer

L Raymond-King, S McGrath, B Mukherjee, Y Wang, H Parikh, et al.
The Lancet Regional Health – Southeast Asia

The Case for an Integrated Biobanking Initiative in South Asia

M Salvatore, Y Wang, P Syal, B Wahl, H Parikh, et al.
JAMA Health Forum

Lasting Income Costs of Mental and Physical Illness

EK Johnson, H Parikh, CK Ettman, G Ge, L Sopina, AY Chang
JAMA Health Forum

Ten Core Concepts for Ensuring Data Equity in Public Health

Y Wang, AE Boyd, L Rountree, Y Ren, K Nyhan, R Nagar, J Higginbottom, ML Ranney, H Parikh, B Mukherjee
JRSS: Series C

Precision Mental Health: Predicting Heterogeneous Treatment Effects for Depression through Data Integration

CL Brantner, TQ Nguyen, H Parikh, C Zhao, H Hong, EA Stuart
Intelligence-Based Medicine

CICADAS: A Simulation-Based Framework for Randomized Trial Emulation in Critical-Care Seizure Treatment

M McCauley et al. (including H Parikh)
Harvard Data Science Review · provisionally accepted

Demystifying Prediction Powered Inference

Y Song, DM Kluger, H Parikh, T Gu
Canadian Journal of Statistics · accepted

Data (in)equities in Data Science: Dissecting Systemic and Systematic Biases in Pulse Oximetry

L Rountree, H Parikh, B Mukherjee
AISTATS

Regularizing Extrapolation in Causal Inference

D Arbour, H Parikh, B Niknam, EA Stuart, KE Rudolph, A Feller
NeurIPS · accepted

Controllable Generative Sandbox for Causal Inference

Q Zhang, H Parikh, A Naimi, R Nabi, C Kim, T Lash
NeurIPS · accepted

Characterizing Underrepresentation in Generalizing Causal Survival Estimates

B Liu, S McGrath, Y Hou, E Stuart, H Parikh
Preprint · arXiv

Proximal Causal Inference through Cross-Proxy Balancing

GV Ringlein, TQ Nguyen, EA Stuart, H Parikh
Preprint · arXiv

Towards Optimal Estimators for Randomized Control Trials

H Parikh, G Levin-Konigsberg, N Tripuraneni, D Madeka, MI Jordan, D Foster, D Perrault-Joncas, A Volfovsky
Preprint · arXiv

Mind the Sim-to-Real Gap & Think Like a Scientist

H Parikh, G Levin-Konigsberg, D Perrault-Joncas, A Volfovsky
Preprint · arXiv

TEA-Time: Transporting Effects Across Time

H Parikh, G Levin-Konigsberg, D Perrault-Joncas, A Volfovsky
Preprint · arXiv

When Does Agroforestry Income Reduce Deforestation? Evidence from a Natural Experiment in Madagascar

C DeSisto, R Rasolofoson, M Foley, H Parikh
Preprint · arXiv

Why Are There Many Equally Good Models? An Anatomy of the Rashomon Effect

H Parikh

2025

Journal of the American Statistical Association

Who Are We Missing? A Principled Approach to Characterizing the Underrepresented Population

H Parikh, RK Ross, EA Stuart, KE Rudolph
Nature Communications

Breast Cancer and Income Loss in Denmark: Heterogeneous Outcomes and Longitudinal Effects

EK Johnson, H Parikh, KR Olsen, AY Chang, L Sopina
International Journal of Radiation Oncology, Biology, Physics

Representativeness and Generalizability of NCI-Funded Multi-Modality RCTs

CFPM de Sousa, H Parikh, JD Bradley, E Stuart, C Hu
AISTATS

Graph Machine Learning based Doubly Robust Estimator for Network Causal Effects

SB Khatami, H Parikh, H Chen, S Roy, B Salimi
NeurIPS

A Cautionary Tale on Integrating Studies with Disparate Outcome Measures

H Parikh, TQ Nguyen, EA Stuart, KE Rudolph, CH Miles
NeurIPS

Data Fusion for Partial Identification of Causal Effects

Q Lanners, C Rudin, A Volfovsky, H Parikh
Preprint · arXiv

Demystifying Proximal Causal Inference

GV Ringlein, TQ Nguyen, PP Zandi, EA Stuart, H Parikh
Preprint · arXiv

Towards Enhancing Data Equity in Public Health Data Science

Y Wang, AE Boyd, L Rountree, Y Ren, K Nyhan, R Nagar, J Higginbottom, ML Ranney, H Parikh, B Mukherjee

2024

Annals of Clinical and Translational Neurology

How Many Patients Do You Need? Trial Designs for Anti-Seizure Treatment

H Parikh, H Sun, R Amerineni, ES Rosenthal, A Volfovsky, C Rudin, MB Westover, SF Zafar
AISTATS

Safe and Interpretable Estimation of Optimal Treatment Regimes

H Parikh, Q Lanners, Z Akras, S Zafar, MB Westover, C Rudin, A Volfovsky
AISTATS

Interpretable Causal Inference for Wearable, Sensor, and Distributional Data

S Katta, H Parikh, C Rudin, A Volfovsky
Harvard Data Science Review

Towards Generalizing Inferences from Trials to Target Populations

MY Huang, H Parikh

2023

The Lancet Digital Health

Effects of Epileptiform Activity on Discharge Outcome in Critically Ill Patients

H Parikh, K Hoffman, H Sun, SF Zafar, W Ge, J Jing, L Liu, J Sun, A Struck, A Volfovsky, C Rudin, MB Westover
ICLR · Tiny Papers

Synthetic Control as Balancing Scores

H Parikh
UAI

Variable Importance Matching for Causal Inference

Q Lanners, H Parikh, A Volfovsky, C Rudin, D Page

2022

Journal of Machine Learning Research

MALTS: Matching After Learning to Stretch

H Parikh, A Volfovsky, C Rudin
ICML

Validating Causal Inference Methods

H Parikh, C Varjao, L Xu, EJ Tchetgen Tchetgen
NeurIPS · Demonstrations Track

AME: Interpretable Almost Exact Matching for Causal Inference

H Jiang, T Howell, NR Gupta, V Orlandi, M Morucci, H Parikh, S Roy, C Rudin, A Volfovsky

2020–2021

SIGMOD

Causal Relational Learning

B Salimi, H Parikh, M Kayali, L Getoor, S Roy, D Suciu
Preprint · arXiv

Heterogeneous Treatment Effects in Social Networks

A Gilad, H Parikh, S Roy, B Salimi

People

The CEADS community

Members and affiliates of the Causal Evidence and Decisions Studio.

Harsh Parikh
Harsh Parikh
Sean McGrath
Sean McGrath
Lillian Rountree
Lillian Rountree
Emily Johnson
Emily Johnson
Grace Ringlein
Grace Ringlein
Srikar Katta
Srikar Katta
Camille DeSisto
Camille DeSisto
Peter Liu
Peter Liu
Yiren Hou
Yiren Hou
Ana Karina Raygoza Cortez
Ana Karina Raygoza Cortez
Quinn Lanners
Quinn Lanners
Ge Ge
Ge Ge
Qi Zhang
Qi Zhang

Teaching

Learning through practice.

Fall 2026 · Yale

BIS 527: Introduction to Health Data Science

Algorithmic thinking, prediction, and data visualization in public health, with practical work in Python.

Course website

Tutorial

Interpretable machine learning

Methods and principles for interpretable machine learning, with activities using real data.

Explore the tutorial

Contact

Let’s connect.

harsh.parikh@yale.edu
Department of Biostatistics
Yale School of Public Health
New Haven, Connecticut

Prospective students & collaborators

If you are interested in working with CEADS on causal inference, machine learning, or public health, please email me with your CV and a short description of your research interests.

I also maintain a job board for statistics, data science, and AI, with opportunities across institutions.