Particle.news
Download on the App Store
7 ARTICLES
·
last wk.

Tamil Nadu Launches Machine-Learning TB Death Risk Tool Across 2,800 Public Facilities

The integrated feature assigns a probability-of-death score at diagnosis, prompting urgent referrals by shortening admission delays.

Representational image. Doctors demonstrating sputum sample collection steps outside a PHC | Sumi Sukanya Dutta | ThePrint
Image
Image
Image

Overview

  • The machine-learning model went live this week on TB SeWA in all 2,800 public health facilities across Tamil Nadu.
  • It calculates individual mortality risk at diagnosis using five rapid clinical markers, generating probabilities from 1% to 50%.
  • Health workers can now use these scores to prioritize hospital admissions and reduce the one-quarter of severely ill patients who currently wait three to six days.
  • The tool was trained on data from 56,000 adult TB cases collected under Tamil Nadu’s TN-KET framework between July 2022 and June 2023.
  • TN-KET’s differentiated care approach had already reduced early TB deaths by 20% statewide within two quarters and achieved 20–30% declines in two-thirds of districts.