Pharmacist-run drug side-effect reporting for hospitals that file none today
TL;DR Harmlog places a remote pharmacist team and a simple ward tool inside private and district hospitals that send almost no adverse drug reaction reports today. It finds likely reactions in discharge notes and ward logs, completes the national form, and files it through the regional PvPI centre. Hospitals pay a monthly fee per bed.
India's drug-safety system depends on hospitals reporting what they see, and many report nothing. One monitoring centre found that private medical colleges and peripheral, sub-district and district hospitals in its area sent no reports at all[1]. A centre in Karaikal had received nil reports from nearby hospitals for two years, and more than half of the 200 staff it surveyed believed reporting carried legal risk[2].
Patients barely know they can report. Only 6.4% of people surveyed in Madhya Pradesh had heard of the consumer side-effect form[3]. Rare, serious reactions still turn up when data exists. India sent 313 reports of paracetamol-linked Stevens-Johnson syndrome or toxic epidermal necrolysis to the WHO database over 12 years[4].
What it does, in regulatory terms: adverse event case capture and reporting support only. Harmlog doesn't diagnose patients, assess causality on its own authority or change treatment. Each report goes to the national programme through its existing route[5], and the treating doctor confirms every case.
A hospital installs a short ward form and shares discharge summaries. Harmlog's pharmacists screen them for common trigger signs, such as a drug stopped for a rash, an antidote given or an unexplained fall in blood counts. They ask the treating doctor to confirm, complete the national reporting form and file through the regional monitoring centre. The hospital gets a monthly safety summary and a record it can show inspectors.
The seed pitched AI signal detection, and I've reframed it. Signals come from pooled national and global data held by the Pharmacovigilance Programme of India and the WHO centre[5]. A startup can't see that data, but it can fix the intake gap in hospitals that send nothing.
Doctors and nurses use it but the hospital pays. The test price is a monthly fee per bed, which is my assumption until interviews confirm it. Today the alternative is free - a hospital can file paper or app reports itself[5], but in practice nobody owns the job. Mid-sized private hospitals and chains with many smaller units are the first buyers.
Monitoring centres have shown that structured training plus phone reminders raise reporting from silent hospitals[2], which is close to what Harmlog would sell as a service. Consumer reporting forms exist and people are willing to use them once someone tells them. In fact 91.63% said they would fill the form[3].
No public figure exists for spending on hospital pharmacovigilance in India. The narrowest signal is the gap itself - a 2015 study counted 90 operational monitoring centres with a functional rate of 56.45%[1]. Treat that as evidence of under-reporting, not of buyers with budgets.
PvPI's own monitoring centres, mostly in medical colleges, collect reports and offer a mobile app[5]. The UK's Yellow Card scheme lets anyone report problems with medicines and even health apps[6], which shows how simple intake can be. Pharmacovigilance outsourcers serve drug makers, not hospitals. I found no service selling report capture to Indian private hospitals.
| Difficulty | Moderate (39/100) |
| Starting capital | ₹35–70 L |
| Time to launch | 4–6 months |
The software is light. The hard part is the sale, because hospitals may fear that more reports look like more harm and nobody is fined for reporting nothing. You also need pharmacists who can judge cases quickly and doctors who are willing to confirm them.
No licence to operate. You work through the national programme's existing route.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
Don't let AI decide causality. Whether a drug caused a reaction is the treating doctor's judgement, and a wrong call in a national safety report is worse than none.
Want to work out the AI layer for Harmlog? Talk to Yash →
Here's what can go south -
I think this suits a pharmacologist or clinical pharmacist who has worked at a monitoring centre and can sell to hospital administrators. If you want to build signal-detection software then this is the wrong idea. Prove first that five hospitals will pay and that each produces at least ten confirmed reports a quarter.
Plan for roughly ₹35–70 L, mostly pharmacists and the slow sale to hospital administrators.
No. Reports go through PvPI's monitoring centres on the national form. You need DPDP processor terms with each hospital, and doctors confirm every case.
Yes. Most countries run a reporting scheme like the UK's Yellow Card. Check how your regulator accepts reports from third parties first.
The free Harmlog validation kit has the line-by-line capital breakdown, the effort scoring, the pharmacovigilance and data-protection checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for hospitals, and 5 assumptions to test. Download the kit →
Thinking about building Harmlog or something close to it? The make-or-break work is persuading hospital leaders that reporting protects them rather than exposing them. If you need a hand with that pitch, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.