Context:
The PM’s ambition of eradicating TB by 2025 is stymied due to lackluster bureaucratic action.
TB screening AI chest X ray tools have been evaluated and approved by health agencies but are not being used in any capacity by the national programme.
Central TB Division (CTD) of the Health Ministry is supporting an unverified AI tool DeepCXR while neglecting two highly researched, effective, and affordably priced avenues qXR and Genki.
AI Options for TB Screening
- Two AI technologies qXR (Qure.ai, Bengaluru) and Genki (DeepTek, Pune):
- Approved by HTA and Medical Technology Assessment Board (MTAB).
- Mostly tested and validated with published studies.
- Implementation in Over 3,100 sites across 90 countries (qXR) and in 80 sites across 15 Indian States (Genki).
- Highly Cost effective at ₹30 per scan (qXR) and ₹22 per scan (Genki).
- Sensitivity: Really over 90%, keeping in line with WHO TB screening accuracy standards.
DeepCXR (Institute for Plasma Research, Gandhinagar)
- Recommended for the national TB programme, however, it lacks peer reviewed validation.
- No published data regarding its field performance, sensitivity or specificity.
- DeepCXR is trained on 54,000 X rays, however, it is uncertain whether it has ever been used in real world TB screening.
Bureaucratic Delays and Inconsistencies
- HTA approval is not essential, but the Central TB Division is waiting for it before proceeding with new TB technologies.
- Notwithstanding their approval, qXR and Genki remain excluded from the national programme.
- Instead, the unvalidated DeepCXR was recommended, with no formal communication to the respective States.
- States became aware of DeepCXR when they inquired into AI solutions for TB detection.
Importance of AI Assisted TB Screening
- X ray screening is fundamental
- 42.6% of the TB cases in India were detected by X ray screening in the National TB Survey conducted from 2019 to 2021.
- 39% of TB cases were detected solely by X ray screening in the 2021 22 survey conducted in Tamil Nadu.
- AI driven interpretation of X rays
- Fast (in under a minute).
- Cost effective.
- Feasible in low resource settings.
The Central TB Division must take urgent action on validated AI tools; that is, qXR and Genki must be rolled out for immediate acceleration in TB detection.
Clear policy guidelines are needed to avoid bureaucratic red tapes.
Ensuring AI solutions undergo independent validation prior to their adoption into national programmes.
