SynthEHRella is a benchmarking package used for evaluating synthetic Electronic Health Records (EHR) data generation methods.

EHR data engineering using MIMIC-IV and K-MIMIC

To transform the MIMIC-IV FHIR dataset into a structured format suitable for analysis by converting its JSON files containing patient, condition, and encounter data into a CSV file that maps patient IDs to condition timestamps. This involves extracting, processing, and merging data to create a comprehensive and analyzable dataset.

Extraction and analysis of data in the MIMIC IV database

Queries needed to recreate time series features for model training

Add the MIMIC-IV data to a DuckDB database

Intended to implement multiple-indicator, multiple-cause (MIMIC) modelling with discrete indicators in R.

MIMIC data importer for PostgreSQL

TemporAI-MIVDP: Adaptation of MIMIC-IV-Data-Pipeline for TemporAI

Implementation of the Transformer with Multi-channel attention mechanism and evalutation on the MIMIC-IV

AIDx Model Development: Developing AI-driven models for healthcare applications.

Towards a CKD (Chronic Kidney Disease) dataset.

Code for paper using causal inference to explore the process of weaning ICU patients off of mechanical ventilation.

Team Cyan's function toolbox for exploring databases

Research Methodology Project

SG Healthcare AI Datathon 2021 - acute kidney injury (AKI) patients requiring replacement renal therapy

Patient Discharge Classification based on the Hospital Treatment Process (ICPM 2021)

Official repo for "Characterizing Stigmatizing Language in Medical Records" (ACL 2023)

Embeddings generation from MIMIC-IV and MIMIC-CXR

MIMICAI - Exploring MIMIC Data Using LLM and Open-WebUI