machine learning Archive — Page 1 of 2

July 2, 2026

Team’s prediction task compares GPT-4o with classic machine learning

Large language models have been functionally opaque. Seeking some transparency, a team undertook a comparison with traditional machine learning for predicting which patients would discontinue their home cancer medications.

June 17, 2026

Team’s post-op kidney injury risk model could aid prevention

The machine learning-derived model was 88% accurate in ranking preoperative patients according to risk for postoperative acute kidney injury.

May 29, 2026

AI technique improves cancer gene discovery for breast and prostate cancer

The artificial intelligence model Enformer was retrained with tissue-specific datasets; this “transfer learning” approach outperformed the base model in identifying disease-associated genes.

An AI-driven rendering and analysis of a holotomography-based 3D reconstruction of a tumor tissue sample derived from a spatial molecular experimental platform. Cancer cells (blue), stromal cells (green), and immune cells (red) are highlighted, while the extracellular matrix (ECM) appears in translucent green with a green arrow indicating its orientation. By integrating spatial molecular measurements, this approach offers single-cell and subcellular functional characterization, providing key insights into cellular interactions and communication in a three-dimensional context to deepen our understanding of the tissue’s architecture and composition.
January 16, 2025

VUMC to launch Molecular AI initiative to spur precision medicine, transplantation

The leader of the new initiative, Tae Hyun Hwang, PhD, will work to apply AI and other revolutionary technologies, including advanced molecular imaging techniques, to clinical practice.

January 3, 2025

AI tested for alerting clinicians of suicide risk at three VUMC clinics

At neurology clinics the system was used to flag 8% of arriving patients as having relatively high risk for suicide attempt in the next 30 days.

June 24, 2024

Automated algorithm predicts risk of blood clots in hospitalized patients

This new entry to the field works in the background to provide real-time risk assessments, with no manual inputs from health care providers required.