Varun Mishra spent two decades keeping digital infrastructure running at LinkedIn, Lazada and ByteDance. Now he says the connected health sector is making the same mistake he learned to avoid on day one.
Start with a number. Missed medication doses cost the NHS over £500 million a year. Not in procurement. Not in administration. In avoidable readmissions and complications that happen because patients did not take the right drug at the right time. The technology to track this exists. It has existed for years. So why is the number still £500 million?
Varun Mishra thinks he knows the answer, and it has nothing to do with the devices themselves.
Mishra has spent his career building infrastructure that cannot afford to fail. Five years at LinkedIn, managing data centre systems for a platform with more than 600 million users. Four years before that at Lazada, the Alibaba-backed e-commerce giant, where he rose from Senior Systems Engineer to Manager and built automated infrastructure across six countries serving tens of millions of active buyers. He now runs technical programme management for ByteDance across EMEA. In mid-2024, alongside that role, he co-founded Adhicine Solutions Ltd, a UK company working on connected medication adherence devices for the NHS.
The problem he is going after is not a hardware problem or a software problem. It is an assumption problem.
Built for the Wrong Patient
Most connected health devices work like this: the device collects data and sends it to a cloud server, the server processes it, and alerts go back to the clinical team. Clean, logical, sensible. As long as the internet connection holds.
In a hospital in central London, it holds. In a care home in rural Lincolnshire, or a district clinic two hours outside Bangalore, it often does not. And when it drops, the device does not slow down. It stops doing its job entirely.
“The fundamental assumption built into most connected health products is that the network is always available. Deploy those products in a rural clinic, a care home, or a smaller city facility, and you do not get a slower product. You get one that does not function at the moments the patient needs it most.”
Varun Mishra, Co-Founder and CTO, Adhicine Solutions Ltd
Mishra is careful not to frame this as a criticism of the companies building these products. Most of them are doing exactly what makes commercial sense: design for the environment where the device is easiest to test and demonstrate. That environment tends to be urban, well-connected and nothing like the settings where chronic disease is most prevalent.
India makes the point starkly. The National Digital Health Mission is building national infrastructure to connect health records, prescriptions and monitoring devices for 1.4 billion people. The chronic conditions it is targeting — diabetes, cardiovascular disease, tuberculosis — fall disproportionately on rural and peri-urban populations. The devices feeding into this infrastructure were mostly tested somewhere else.
“The direction of India’s digital health mission is right and the problem it is tackling is genuinely important. But the engineering culture in the sector has to shift. Stop treating connectivity as a given and start treating it as a variable. When you design for the worst connection you will encounter rather than the best, the product works where it is most needed, not just where it looks good.”
Varun Mishra
What He Learned at Lazada
The fix Adhicine is building inverts the standard architecture. Instead of running intelligence in the cloud and pushing results to the device, the device runs the intelligence itself. A machine learning model sits on the hardware, builds a picture of each patient’s dosing patterns over time, and generates alerts locally. No live connection required. When connectivity comes back, the device syncs. The patient never notices the gap.
This is not a novel engineering concept. It is called edge computing, and it has been standard practice in commercial infrastructure for years. The insight Mishra brings is that connected health has been slow to adopt it, and that the reason is cultural as much as technical.
“At Lazada we were building across six countries with very different infrastructure realities. The engineering lesson was simple: you cannot treat connectivity as a constant. Either the product works without it, or it fails the people who need it most. For a shopping platform that means a lost sale. For a medication adherence device it means a missed dose.”
Varun Mishra
Running systems for hundreds of millions of users teaches a specific discipline: plan for what breaks, not for what works. The failure modes that matter are not the ones you can reproduce in a test environment. They are the ones that appear in production, at two in the morning, in infrastructure you do not control.
Early Days, but the Argument Is Bigger Than One Product
Adhicine is still in development. Mishra is clear about that. There is no product on shelves, no clinical trial data, no NHS contract yet. What there is, is a specific technical argument about how the category is being built, made by someone who has spent two decades at the coalface of infrastructure engineering.
That argument does not rise or fall with Adhicine. The NHS Long Term Plan, India’s NDHM and comparable programmes across Southeast Asia and Sub-Saharan Africa are collectively deploying billions in connected health infrastructure right now, mostly on the same cloud-first assumptions Mishra is questioning. Whether those deployments succeed in reaching the patient populations they are designed for may depend less on the quality of the devices than on a decision most vendors are not yet being asked to make.
Build for the best connection, and you build for the easiest patient. Build for the worst, and you might actually reach the rest.
Last updated: June 11, 2026
