Rishi Lamichhane

Mapping Tobacco Control and Shared Services Funding in Massachusetts

Collected and cleaned state grant data on tobacco control and shared public health services, then mapped it across Massachusetts communities in ArcGIS. Shared with the Massachusetts Health Officers Association for official and public use. (Aug–Sep 2023)

Breast Cancer Classification with Machine Learning

Can cell measurements from a needle biopsy predict whether a breast tumor is benign or malignant? Using the Wisconsin Diagnostic dataset, I compared logistic regression, a decision tree, k-nearest neighbors, and a neural network with 5-fold cross-validation. Logistic regression, KNN, and the neural network each reached about 95% test accuracy. (Computer Systems Seminar, Spring 2024)

Evaluating Public Health Trainings at the MHOA Annual Conference

Turned 1,828 raw evaluation forms from a 3,682-attendee conference into session-level summaries in Excel. Comparing ratings across 26 trainings showed attendees rated instructors more highly than the knowledge they gained in 25 of 26 sessions, pointing to a need for more practical, job-applicable content. (Massachusetts Health Officers Association, 2023)

Research project · NourishED Research Foundation, with Dr. Brenna Bray

Network Mapping of Qualitative Data: Comparing Kumu and Python

Built two network-mapping protocols, Kumu and Python (NetworkX), to visualize how themes from a qualitative study of expert perspectives on binge eating disorder connect, and used centrality metrics such as betweenness and eigenvector to identify the core themes. Invalidating environments, systemic oppression, and marginalized populations sat at the center of the network. Lead author on the manuscript in preparation.

Community health project · Self-initiated and self-funded

Teaching Tobacco Risk Through Hands-On Learning in Nepal

Designed and led tobacco-prevention sessions for students aged 13–17 in government schools in Kathmandu. Using an experiential activity, students breathed through paper tubes of different widths to feel the difference between healthy and smoking-damaged lungs, turning long-term health risk into something they could experience directly. (May–Jun 2023)

RI-AI4H Datathon · Brown Center for Biomedical Informatics · Top 7 · with Rushil

CardioPALS: A Machine Learning CVD Risk Calculator for Rhode Island

Built a predict-and-prevent framework for cardiovascular disease using 16 clinical, behavioral, demographic, and geographic risk factors, versus 6–8 in standard scores like Framingham. XGBoost reached an AUC of 0.95 (logistic regression: 0.90) on synthetic Rhode Island patient data, and SHAP values identified each patient's top risk drivers to match them with evidence-based interventions. (July 2026)

Independent project · R

Mapping the HEOR Field: Academia–Industry Collaboration in Value in Health

Pulled all 1,071 research articles published in Value in Health from 2021 to 2025 via the PubMed API, classified 6,762 author affiliations by sector, and built a co-authorship network. About 1 in 6 papers was an academic–industry collaboration, a share that held steady over five years. The field's main bridges, measured by betweenness centrality, were academic health economics units and HTA bodies such as NICE, with none from industry.