Ongoing Projects

Joint Initiative for Causal Inference

Sponsor/Partner: Novo Nordisk

Project Description/Goals: To create an international powerhouse for statistical methods within casual inference to be used on RCT and observational data with a hub at Copenhagen University as well as at University of California, Berkeley by developing, implementing and disseminating methods for exploiting vast, new health datasets using state-of-the art advances in machine learning, causal inference, and statistical theory, and to build industry-wide consensus around best practices for answering pressing health questions in the modern methodological and data ecosystem.

CTML Faculty Involved: Maya Petersen M.D. Ph.D., Mark van der Laan Ph.D., Laura Balzer PhD., and Andrew Mertens PhD.

Gilead-Berkeley Global Health Equity Initiative

Sponsor/Partner:   Gilead Sciences

Project Description/Goals: The UC Berkeley School of Public Health and Gilead Sciences have launched the Gilead-Berkeley Global Health Equity Initiative to address real-world public health issues. Funded at $4.5 million for three years, the program focuses on infectious and non-communicable diseases and aims to enhance research, data management, and executive education. The initiative has three components: collaborations in applied research, involving doctoral students and junior faculty at the Center for Global Health; collaborations in biostatistics and data management under the CTML - Center for Targeted Machine Learning and Causal Inference and executive education.

 CTML Faculty Involved: Alan Hubbard Ph.D., Maya Petersen M.D. Ph.D., Mark van der Laan Ph.D., Laura Balzer Ph.D., Alejandro Schuler Ph.D., and Andrew Mertens PhD.

Effects of a State-Wide Medicaid Risk Tiering Algorithm on Spending & Population Health

Sponsor/Partner: Arnold Venture Foundation

Project Description/Goals: In 2025, Medi-Cal became the first state Medicaid program to roll out its own statewide risk tiering system built on machine learning. The Risk Stratification, Segmentation, and Tiering (RSST) algorithm, developed by the California Department of Health Care Services (DHCS), sorts roughly ten million members into high-, medium-, and low-risk tiers each month. Most risk tools focus on cost and hospital use. Working with DHCS, the team from CTML, Tufts, Yale, and Brown will now build a randomized evaluation into the statewide rollout. The study will measure whether a high-risk flag, and the care that follows it, prevents adverse outcomes and changes Medicaid spending over the next year. It will also test whether members below the high-risk cutoff would benefit from a flag. The findings will guide how California refines RSST and give other states rigorous evidence as they consider similar tools.

 CTML Faculty Involved: Alejandro Schuler Ph.D. and Maya Petersen M.D. Ph.D.

SEARCH Community Precision Health Study

Sponsor/Partner: NIH National Institute of Health  (Subaward through UCSF)

Project Description/Goals: The Sustainable East Africa Research in Community Health (SEARCH) consortium designs and tests evidence based treatment and prevention interventions that will guide a global effort to end AIDS using a multi-disease, multi-sector approach.  

CTML Faculty Involved: Maya Petersen M.D. Ph.D. and Laura Balzer Ph.D.

Toxic Substances in the Environment (Superfund)

Sponsor/Partner:  NIEHS

Project Description/Goals:   Identify problems associated with hazardous waste sites that have proven intractable to current methods, applying a new approach called ‘exposomics’ to address these problems and employ novel cutting-edge scientific methods and direct engagement with key stakeholders and communities in a concerted effort to make a public health impact.

 CTML Faculty Involved: Alan Hubbard Ph.D.

Targeted Machine Learning to evaluate and optimize HIV prevention strategies in cluster randomized trials

Sponsor/Partner: NIH National Institute of Mental Health

Project Description/Goals: Cluster-randomized trials evaluate interventions implemented at the community, clinic, or regional level rather than the individual level. While powerful for measuring population-level and spillover effects (like reduced contagion or social interaction effects), traditional analyses face major hurdles. The project address critical statistical and analytical challenges when evaluating cluster-randomized trials (CRTs) for HIV prevention (such as universal test-and-treat, PrEP delivery, or community-level structural interventions).

CTML Faculty Involved: Laura Balzer PhD 

Toxic Peer-led Dynamic Choice HIV Prevention for Women at Elevated HIV Risk in Uganda (subaward through UC San Francisco)

Sponsor/Partner: NIH National Institute of Mental Health

Project Description/Goals: This study is in two phases. In Aim 1, we refined and pilot tested the Peer-led Dynamic Choice HIV Prevention (DCP) intervention in a single-arm pilot trial in Ibanda district to test the hypothesis that the Peer-led Dynamic Choice HIV Prevention intervention will be feasible, acceptable, and increase biomedical HIV prevention coverage in the 6 months after receipt of the Peer-led DCP intervention compared to the standard of care in the 6 months prior to intervention implementation. Aim 2 is a Hybrid Type 1 cluster-randomised trial to test the Peer-led DCP intervention against the standard of care. The trial will include 14 clusters, each within the catchment area of a government health centre that offers standard-of-care HIV prevention services. The trial population will include approximately 560 HIV-noninfected women with a PrEP indication or at elevated risk of HIV exposure followed for up to 96 weeks.

CTML Faculty Involved: Laura Balzer PhD

Toxic effects of a State-Wide Medicaid Risk Tiering Algorithm on Spending & Population Health

Sponsor/Partner: Laura and John Arnold Foundation

Project Description/Goals: Estimating the impact of the Medi-Cal algorithm, a machine learning-based risk stratification system used by the California Department of Health Care Services, on health outcomes, health care utilization, and spending by conducting a randomized controlled trial (“RCT”).

CTML Faculty Involved: Alejandro Schuler PhD and Maya Petersen PhD

Estimating global, national, sub-national prevalence of Vitamin and Mineral Deficiencies (VMD) using machine learning models

Sponsor/Partner: Bill & Melinda Gates Foundation (subaward through UC Davis)

Project Description/Goals: Develop innovative geospatially-explicit machine learning models to estimate the prevalence of Vitamin and Mineral Deficiencies (VMD) at global, national, and sub-national levels. The models that will leverage widely available data that are plausibly associated with the micronutrient deficiencies, such as food availability, dietary diversity indicators, anthropometric indicators (stunting, wasting, underweight), prevalence of specific infections, socio-economic indicators, and existing micronutrient deficiency control programs, among other potential predictors of deficiency. The resulting proxies will be validated against deficiency estimates based on the micronutrient status biomarkers for micronutrients of concern, obtained from nationally representative surveys

CTML Faculty Involved: Andrew Mertens PhD

Expanding Safe Water Access to Improve Health Outcomes in Appalachia: The Rural Water Filtration and Health (RWELL) Trial (subaward to Virginia Polytechnic Institute)

Sponsor/Partner: NIH National Institute of Allergy & Infectious Diseases

Project Description/Goals: To improve health outcomes in the Central Appalachia area by qualitative and quantitative assessment focused on the sustained use of free countertop water filters to lower-income households with private well and spring water in rural Virginia and Tennessee counties.

CTML Faculty Involved: Alan Hubbard PhD

SMART-SEPSIS: Sequential Machine learning for early Antibiotic individual Response Targeting in suspected Sepsis (subaward to UC San Francisco)

Sponsor/Partner: NIH National Heart, Lung, and Blood Institute

Project Description/Goals: The core mission of SMART-SEPSIS is to transition from uniform treatment mandates to individualized precision medicine at the bedside. It accomplishes this by using sequential machine learning models to analyze real-time, longitudinal patient data from Electronic Health Records (EHRs).

CTML Faculty Involved: Alan Hubbard PhD

The effects of air pollution and temperature on perinatal outcomes

Sponsor/Partner: Bill & Melinda Gates Foundation (subaward to UC San Francisco)

Project Description/Goals: The project investigates how environmental stressors—specifically ambient air pollution and temperature fluctuations—impact perinatal outcomes (such as birth weight, gestational age, and infant survival/health).

CTML Faculty Involved: Alan Hubbard PhD

Inside StART: Implementation of Novel Strategies for long-acting ART (subaward to UC San Francisco)

Sponsor/Partner: NIH National Institute of Allergy & Infectious Diseases

Project Description/Goals: Globally 1.9 million stillbirths and 2.3 million neonatal deaths occurred in 2021. A disproportionate number of these outcomes occur in South Asia, predominantly in India where almost 350,000 stillbirths and over 400,000 neonatal deaths occur annually. The project aims to quatify the national burden of stillbirth and perinatal mortality in India attributable to ambient PM2.5 and temperature, the Population Attributable Fraction (PAF), and Disability Adjusted Life Years (DALY) by examining the association of ambient exposure to PM2.5 and temperature during pregnancy with stillbirth and perinatal mortality in India, between 2014 and 2021, using data from the National Family Health Survey (NFHS) 5.

CTML Faculty Involved: Alan Hubbard PhD

This work by CTML Faculty and/or Staff is licensed under CC BY 4.0