ML-in-healtchare

Repository containing code corresponding to the lab assignments and final project of the Machine learning in healthcare course. Fall 2023.

Lab 1

Implementation of an MVAE (multi-view variational autoencoder), capable of learning meaningful, disentangled latent representations from multiple image views. The approach implemented supports image processing however it can be easily extended to multi-modal data through contextual embeddings.

Lab 2

Complete survival analysis project using Kaplan-Meier and Cox-Hazard models on latent representations of data obtained through UMAP based on subgroups found through mixture models maximizing BIC.

Main code is located inside the notebook

Final project

• Title: Unsupervised analysis of the relationship between various genetic polymorphisms and their relationship to impulse-aggresiveness personality traits

1. Dimensionality Reduction and Clustering:

   • Utilize statistical criteria for dimensionality reduction and clustering to create patient profiles.
   • Enhance our understanding of the patient population by identifying patterns in the data.

2. Frequent Itemsets of Polymorphisms for Personality Types:

   • Analyze frequent itemsets of genetic polymorphisms associated with various personality types.
   • Considering the limitations of imputed data in clinical value, focus on polymorphisms of different lengths.
   • Contribute to the comprehension of DNA mutation distribution and its impact on patient personality.

3. Feature Importance Analysis:

   • Employ a supervised learning approach using a random forest to determine feature importances.
   • Identify the key features that play a significant role in determining personality types.
   • Enhance interpretability and insights for clinical decision-making.

4. Survival Analysis based on Polymorphisms:

   • Implement survival analysis methods after dimensionality reduction and clustering.
   • Explore the impact of genetic polymorphisms on the likelihood of suicidal ideation.
   • Improve our understanding of the long-term effects of these polymorphisms and their potential clinical implications.