Add us as a preferred source
Kamya Elawadhi, co-founder and president of Doceree explains why real-time intelligence at the point of prescribing is important.
The healthcare industry has become remarkably good at collecting data. Every patient interaction, prescription, diagnosis, lab result and reimbursement generates another signal that promises greater insight into clinical decision-making. Yet despite this abundance of information, one challenge remains surprisingly unresolved: delivering the right intelligence while a prescribing decision is actually being made.
Too often, what is labelled as “real-time” intelligence is anything but. It arrives after the prescription has been written, after the patient has left the consultation, or after the opportunity to influence care has passed. At that point, the data may explain what happened, but it cannot change the outcome. As prescribing becomes more complex – with evolving formularies, affordability concerns, and increasing administrative burden – the timing of intelligence has become just as important as the intelligence itself.
The cost of arriving late
Prescribing has become harder, not easier. Formularies shift quarterly. Prior authorisation requirements multiply – physicians now complete an average of 40 requests per week, according to the American Medical Association’s most recent physician survey, consuming roughly 13 hours of physician and staff time that used to go toward patients. Biosimilars and follow-on therapies have added real clinical nuance to decisions that used to be simpler. Physicians are asked to hold more variables in their heads, in less time, than at almost any point I can remember in this field. Meanwhile, patients are more cost-sensitive and more likely to abandon a fill if the first experience – affordability, access, understanding – goes wrong: one peer-reviewed study, tracking abandonment of an HIV prevention medication, found rates of 5.5% at no out-of-pocket cost climbing past 40% once that cost exceeded $500 (£372), with even a small increase from $0 to $10 doubling the abandonment rate, a pattern that shows up across cost-sensitive prescriptions more broadly.
In that environment, a prediction that’s directionally right but temporally late doesn’t help the patient standing at the pharmacy counter deciding whether they can afford what’s in front of them. It helps with a report. The cost of that gap isn’t abstract; it shows up as delayed interventions, missed affordability support, avoidable non-adherence, and, ultimately, worse outcomes that we then spend enormous energy trying to explain retrospectively.
A shift already underway
To be fair, the industry is not standing still on this. There is real momentum toward event-driven, interoperable infrastructure – electronic health record (EHR) systems that expose clinical events as they happen rather than in batch, pharmacy platforms that can signal a fill or an abandonment in near real time, data exchange standards that are finally making real-time a technical possibility rather than a marketing phrase. This is the quiet, unglamorous infrastructure work that rarely gets a headline, but it’s the precondition for everything else.
The distinction I’d draw is between predicted windows and triggered moments. A predicted window says: based on patterns, this physician will likely prescribe something in category X sometime this month. A triggered moment says: this diagnosis was just coded, this script was just written, this fill just happened – respond now, while it’s still relevant. The difference sounds subtle. In practice, it’s the difference between intelligence that’s useful and intelligence that’s interesting. Predictions still matter – they’re how you plan. But a plan is not a presence.
Getting to triggered moments requires a few things that are easy to state and hard to build: integration close to the actual source of the clinical signal, not several steps removed from it; first-party context rather than inferred or look-alike data; and compliance and privacy safeguards designed into the architecture from the start, not layered on afterwards. None of this is exotic. It’s just genuinely difficult infrastructure, and difficult infrastructure is usually where the real value in healthcare technology tends to live.
A moment no one owns
What I find most interesting, though, isn’t the technology. It’s the question of ownership. Who is actually responsible for the point of prescribing as a real-time moment? The EHR vendor has the workflow, but not always the downstream signal. The pharmacy system has the fill data, but rarely the clinical context that preceded it. The payer has the cost picture but sees it weeks later. Marketers and life sciences companies want to be useful in that moment, but are often the furthest from it.
I don’t think any single stakeholder owns this moment today – and none of the incumbent systems was built to. The prescribing journey is fragmented by design – diagnosis, prescribing, dispensing and refill all sit in different systems, run by different companies, on different timelines. Solving for real-time intelligence at the point of prescribing isn’t really a product problem. It’s a connective one, an argument for infrastructure that spans those handoffs rather than infrastructure that optimises any single link in the chain.
That, to me, is the real frontier. Not another dashboard, not another model trained on last month’s data, but genuine presence at the moment decisions get made. Healthcare has spent years getting better at explaining what happened. The next competitive advantage for patients as much as for the industry will belong to whoever gets serious about being present for what’s happening now.
