EPRs are one of the greatest obstacles to preventative medicine

It is not enough to capture past events. Electronic patient records (EPRs) must enable forward-looking precision health, writes Professor Andy Hardy, chief executive of University Hospitals Coventry and Warwickshire NHS Trust

Most EPRs have been designed to capture what has happened in the past, rather than look forward. Precision health – the ability to track an individual’s risk profile over time and adapt care plans accordingly – could help to overcome this weakness.

However, this will require EPRs to process multi-source risk information. If this doesn’t happen EPRs will be one of the biggest obstacles to enabling personalised and preventative medicine in the future.

Precision health uses clinical, genomic, behavioural, environmental and social information to predict and prevent ill health before it shows up in the patient.

The type of information that will have to be communicated through the user interface and workflow of the EPR system is risk patterns and futures, as opposed to just past problems and medications.

The role of the EPR must be as a learning risk platform and not as a digital filing cabinet

This approach has significant implications for the design of the digital infrastructure.

A system designed to manage the storage of encounter-based documentation and a list of problems does not allow healthcare to manage the storage of dynamic risk profiles, model-related insights at the point of care, and support patient-shared decision-making around risk over time.

In the field of precision health, the role of the EPR must be as a learning risk platform and not as a digital filing cabinet.

Engine for risk-informed care

Within the UK and beyond, the focus of the majority of the EPRs has been the standardisation of the information workflow, rather than the impact upon the delivery of care.

The national initiatives in the digitisation of the frontline, as well as the development of the single patient record, have been based upon the principles of avoiding duplication, ease of access, simple decision support and the development of secondary data.

These are the necessary building blocks, but they are insufficient for the requirements of precision health.

Today’s EPR systems remain focused on encounter-oriented charts, often have manual data entry and fixed forms, with decision support that relies on rules versus learning decision support systems that are fuelled by real-world data.

The result is that EPR systems all too often become the final repository for data that has been created elsewhere, rather than the engine that drives real-time, risk-informed care.  

The gold standard for EPR’s efficacy is set by precision health.

Documentation to prediction

EPRs must be able to deal with high-dimensional data – genomic data, imaging biomarkers, digital phenotype data from wearables and social determinants – and deliver this data to the models running inside regulatory environments for decision- making purposes.

This also needs to deal with the data model interoperability and the ability to update the model when new evidence emerges.

From hospital walls to continuous ecosystems

A lot of the data that will enable precision health to be so powerful will never be generated inside the walls of the hospital, but rather inside the domains of community organisations, genomic centres and consumer technology.

EPR systems all too often become the final repository for data that has been created elsewhere

The EPR needs to be designed as a node within a trusted infrastructure, supporting exchange, linkage and federated analysis, as opposed to a monolith that insists that all data must reside within their four walls.

From opaque to governed AI

Because EPRs will contain host AI risk scores and recommendations, it is important that they address issues of explainability, monitoring for bias and accountability to the clinician.

This is far more complex and nuanced than the alerting system that currently exists and will require vision and leadership that sees explainability as a patient safety mandate.

We must also refrain from procuring EPRs that do not support precision health.

Ten years from now, trusts may be layering expensive, piecemeal precision health solutions on top of existing systems which were never designed to support such applications.

They will be struggling with the same issues of integration and safety which the EPR systems were meant to address.

Instead, the customer needs to demand the capability for precision health as a ‘red line’ test. If a platform fails to show scalable solutions for handling multi-source data integration, governance-based AI and risk analysis, it should not progress past the shortlist.

What needs to change 

There are three main transitions that would assist in the implementation of the agenda from rhetoric to reality.

1. A list of procurement criteria that take into consideration the cost of the future. Business cases must be able to place a monetary value on the role of the EPR in enabling precision health use cases over the lifetime of the contract, for example, through open APIs, genomic and image integration, as well as AI.
2. Clinical leadership with an appreciation of the value of data as a clinical asset. High-quality precision health data, complete and well-connected, will not be achieved if the experience of using the electronic patient record by health care practitioners remains solely one of administrative hassle.
3. Regulation and assurance of explainability of reward. A level of regulation and assurance of explainability of reward should be required in relation to the predictive models that become evident through the utilisation of EPRs.

The EPR has always been considered the backbone of digital healthcare. Precision medicine can give it the strength to support a new age of personalised and preventative medicine.

It is now up to healthcare systems to determine what they want to get out of their EPRs. It is no longer in the realm of vendors.

Professor Hardy is a member of the Global Impact Committee which recently published a report into precision health.

He will be speaking at Digital Health Rewired which takes place at the NEC Birmingham on 24-25 March 2026. Register here.