evidoria

← Back to browse

Good practice Imported

ICMR's Hospital Network Adds Machine-Learning Early-Warning Signals to India's National AMR Surveillance

India · New Delhi · See the India profile · See the New Delhi profile

Evidence: Observational / pre–post Top 12% 80/100 · Ask Evidence Copilot about this practice

India's ICMR-run network of tertiary hospitals turned six years of resistance data into machine-learning early-warning signals, flagging cefotaxime resistance as a leading indicator of coming carbapenem resistance across 81,265 bloodstream-infection records from 21 centres.

81,265
Bloodstream-infection records analysed (2017-2022)
31
Tertiary hospital sites in network (2021)
ICMR's Hospital Network Adds Machine-Learning Early-Warning Signals to India's National AMR Surveillance

Details

Maturity
Established
Promoter
Indian Council of Medical Research (ICMR) — AMR Surveillance Network
Period
2016-2024
Keywords
public health, antimicrobial resistance, national surveillance, machine learning, WHO GLASS

Context

Since 2013, India's ICMR has run the AMR Surveillance Network (AMRSN), later supported by the web-based i-AMRSS platform piloted from 2016, for standardised antimicrobial-susceptibility data collection across tertiary hospitals. By 2021 the network spanned 31 sites and had logged more than 280,000 patient records covering 55 antibiotics and antifungals against 116 organisms.

Activities

In 2024, researchers used six years (2017-2022) of bloodstream-infection data from 21 of the network's tertiary centres — 81,265 records — applying time-series analysis to detect lead/lag relationships between resistance trends and k-means clustering to group hospitals and pathogens by resistance pattern.

Results

The analysis identified 'indicator antibiotics' such as cefotaxime, whose rising resistance served as an early-warning signal for resistance to more critical drugs like carbapenems, supporting national antimicrobial-use guidelines aligned with WHO's GLASS standards.

Conclusions

A mature, national-scale surveillance backbone with genuine analytic depth — strong on scale and replication across sites, still maturing on automated validation and open public data access.

Implementation
Indicative cost
Medium (€50k–€500k) — Existing national surveillance network (AMRSN, running since 2013) with an added analytics layer; incremental cost of the ML analysis itself likely modest relative to base surveillance infrastructure.
Time to results
Long (> 3 years) — Network built 2013-2021 (31 sites); ML early-warning analysis published 2024 using 2017-2022 data.
Staffing & skills
hospital laboratory & infection-control staff at 21-31 tertiary centres, ICMR AMRSN central coordination team, data scientists for time-series/clustering analysis

Conditions for success

  • standardised antimicrobial-susceptibility testing protocols across sites
  • sustained multi-year data collection (6+ years)
  • alignment with WHO GLASS reporting standards

Common failure modes

  • automated validation module still not implemented as of 2021 review
  • no link yet to actual antibiotic-consumption data

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful