Digital use cases and real-world data along the patient pathway in patients with hypertension
The goal is to shorten the time from diagnosis to effective therapy – measurably, for every patient group. Digital use cases are the means to that end, not the end itself. As a steering basis for targeted interventions, real-world data is worth considerably more than aggregate statistics – provided the business function, medical, data and regulatory affairs work from the same data basis, together with providers and payers.
Population studies with measured blood pressure and international reports compare actual with documented care. They establish the gaps – as averages. For targeted interventions that is not enough.
of adults aged 30 to 79 in Germany have high blood pressure – roughly 20 million people.2
Why this matters: elevated blood pressure is held responsible for around 54 % of all strokes and 47 % of all cases of ischaemic heart disease.1 Behind the percentage points are avoidable strokes and heart attacks – and people who go without the right therapy longer than necessary.
For context: in the international comparison, Germany ranks among the countries with the highest treatment and control rates.4 So even well-performing systems reach only about half of those affected – the gap is not a German peculiarity but a structural one.
What the average hides. An average says a gap exists. Linked treatment, laboratory and mortality data say where it arises, whom it affects and what it costs: real treatment trajectories, therapy switches, time to blood pressure control, and the link to events and mortality – longitudinal and by cohort instead of a national mean. Only then can an intervention be sized: how large is the addressable group, at which stage of the pathway does care break down, and what effect is realistic?
How wide that variation is shows in an analysis of routine data from 89 English general practices: between the lowest and the highest performing practice there was a more than tenfold difference in the likelihood of achieving the hypertension quality indicator – and most of that variation could not be explained by routinely recorded patient or practice characteristics.6 A German routine-data analysis points the same way: up to 80 % of the variation in quality indicators was attributable to the individual practice – though based on only eight practices.5 Exactly this variation disappears in any average – and with it the entry point for any targeted intervention.
The three use cases improve care at different points of the treatment pathway. Because they act precisely where care actually breaks down, they do not work by watering can: system resources – practice time, diagnostics, avoidable hospitalizations – go where they have the greatest effect and are saved elsewhere. What separates them is how quickly, and with how much effort, they can be delivered.
Linked real-world data – treatment, laboratory and mortality data in one common data model – show, anonymized and close to the process, where, for whom and when care breaks down. Output: a heat map and cohort dashboard instead of an average.
Effect in the pathway: prioritizes and sizes every further measure – and is therefore the precondition for targeting interventions at all.
Risks: data quality and representativeness (panel bias)14, limited data availability in Germany, GDPR requirements, correlation is not causation.
Guideline-aligned recommendations and risk-stratified signatures in the practice system – at the point of the therapy decision.8 In a cluster-randomised trial in Chinese primary care (n = 12,137), a guideline-based CDSS raised guideline-concordant treatment by 15.2 percentage points.7
Effect in the pathway: the strongest substantive lever on quality of care – right where therapy is determined.
Risks: MDR and high-risk AI under the EU AI Act15, liability and reimbursement questions, fragmented system landscape, automation bias. The evidence comes from a different health system and does not transfer unexamined.
Blood pressure self-measurement with escalation logic and patient-facing support close the gap between practice visits – where adherence is actually decided.
Effect in the pathway: keeps patients in the target range and prevents silent drop-off – the stage at which most people are lost.
Risks: MDR, escalation logic as high-risk AI, personal data, selection bias, staffing effort, open reimbursement question.
Each use case is scored from 1 to 5 and weighted. Effect and feasibility carry the most weight; regulatory requirements enter as their own criterion so that they stay visible instead of dominating the discussion.
An initiative of this kind does not begin with a platform, but with a precise data request and a clarified compliance route. Three steps are enough to arrive at a robust basis for decisions.
Define objectives and assessment criteria, agree the clinical questions, formulate the data request and work through the compliance and data protection process.
Build care gap dashboards by region, segment and comorbidity – and check data quality, completeness and representativeness before any interpretation.14
A quantified care gap map, tested success criteria and a decision paper: strengthen the data basis, set up an intervention – or decide against it with reasons.
The success criterion is not the dashboard but a decision: where is an intervention worthwhile, how large is the group affected – and is the data basis good enough to build on?
A care gap map works with anonymized care data without individual-level reference and without a medical purpose: no diagnosis, no monitoring, no therapy recommendation. Processing and storage take place in a European data space, and interpretation of the results stays under human oversight. The legal basis for secondary use is set by the European Health Data Space and Germany’s Health Data Use Act.16
The flip side belongs with it: anonymization limits granularity and rules out any individual outreach. Anyone who wants to address individual patients needs a different route – and a different legal basis. It is precisely this boundary that decides whether an initiative stays quick to deliver or turns into a major regulatory project.
The three use cases are not competing project ideas; they trace a digital roadmap for hypertension care: with each stage, data maturity and clinical effect grow, and each stage creates the data foundation for the next.
Care gap map: quantify the gaps with RWD by region, segment and comorbidity and make the business case robust – fast, and without medical device certification.
Telemonitoring and coaching: use patient-generated data to improve control between practice visits – care steering close to real time.
CDSS and locally trained risk signatures: explainable and source-based, with final sign-off by a human – only meaningful on a solid data foundation.
A larger movement sits behind this: the digital mapping of the care pathway with RWD is becoming increasingly important, and with it evidence generation along the entire pathway – no longer only at the point of approval, but continuously in the reality of care. Care organized this way can improve quality substantially, because it makes effects measurable and directs resources to where they actually change outcomes.
In the large consultancies’ analyses, real-world evidence is moving from a data source to a strategic asset – from downstream evidence to a standing basis for care and portfolio decisions.9, 10 Four forces drive this development; they explain why a care gap map is more than a dashboard project.
Real-world evidence increasingly enters regulatory submissions – for label extensions and safety evidence.10
Payers ask for real-world benefit, not just trial efficacy – under considerable cost pressure.9
Understanding patient groups beyond controlled trials: subpopulations, non-responders, treatment discontinuation.13
Large data volumes become faster and cheaper to analyse – evidence generation becomes scalable.12
How seriously the market takes this shift is visible in the large consultancies' surveys. These figures come from industry surveys and estimation models, not from peer-reviewed research – they describe a direction, not a proven effect:
The right treatment at the right time for the right patient – that is the step from a product-centric to a patient-centric logic.9 Inside a company it does not happen in a single department: the consultancy analyses attribute the success of such efforts explicitly to interdisciplinary teams – clinical medicine, epidemiology, analytics and the business function working together instead of separate analytical cultures.13 The transparency this creates works in both directions: it makes the benefit visible for every function involved – and at the same time shortens the patient’s path to the best possible individual therapy.
The real lever lies in collaboration: when business functions, providers and payers look at the same data basis, tomorrow’s care can be shaped together. Real-world data and continuous data chains are the precondition for that.
The legal framework for this is taking shape: the European Health Data Space and Germany's Health Data Use Act govern the secondary use of health data and will make analyses of exactly this kind more reliably plannable in the coming years.16
Hypertension serves as the example here because the evidence base is solid and the care gap is large. The logic works for any indication and any context in which several AI and digital ideas compete for the same resources: bring clinical benefit, feasibility and regulatory requirements together in one decision paper – and derive a sequence from it instead of a wish list.
That is where my work sits: orchestrating interdisciplinary teams and departments – bringing the business function, medical, data and regulatory affairs to one table, identifying the right use case first, testing it against regulatory reality, and supporting delivery until it is anchored in everyday work.
What is established – and what is not. It is established that real-world data makes care gaps visible, and that a guideline-based CDSS raised guideline adherence in a randomised trial – in a different health system. That data-driven care improves quality overall is plausible and supported by industry analyses, but not established in this form. That distinction belongs in every decision paper – it determines whether an initiative is later measured against its own promises or against what it can actually deliver.
Accessed: August 2026. Sources 1–8 are peer-reviewed or institutional, 9–14 are industry analyses, 15–16 are legal sources.