Medical diagnostics is a process of weighing up information, a task for which computers could be exceptionally well-suited. That argument was made in the 1950s in the journal Science. Today, sixty years later, we are on the eve of the initial breakthroughs, thanks in part to new technological opportunities. Data analytics involves the interplay between medical or other data, data processing techniques (including artificial intelligence and machine learning), patient input, and the professional qualities of health care professionals and data scientists. In data analytics, technology is not the goal in itself, but a means for improving and personalising care. We ultimately seek to achieve the highest possible quality of care for each individual patient. That means that health care issues are among the drivers of the change. To that end, we shall all have to work together in co-creation.
“Splendid isolation won’t do.”
Data analytics alters the diagnostic process
It is my expectation that the diagnostic process will undergo a fundamental change in the coming years. Data and algorithms will play a more governing role. That is due partly to the fact that we ourselves, as individuals, are collecting more and more data at home. Combining that with clinical data from various specialisms, algorithms will arrive at a diagnosis, a probability. On the basis of that diagnosis, a patient, in consultation with a health care provider, will take steps to improve his or her health and quality of life. In other words, the health care professionals themselves will not be rendered redundant by the algorithms.
“Technology will be indispensable as a means of support, but nurses and doctors will still be important for the human aspect in care.”
ADAM Project
Under the name of ADAM – Applied Data Analytics in Medicine – the University Medical Centre in Utrecht (UMCU) has launched a corporate programme that conducts pilot projects. It was initiated and backed in full by the hospital’s Executive Board. Each pilot project is founded not on a research project, but on a concrete clinical question. The aim is to create a digitally supported hospital that provides personalised, data-driven care. In short-cycle pilot projects, multidisciplinary teams work to find innovative solutions for many areas of health care.
“Patients, health care providers and data scientists all sit down together to arrive at a product that is ready for implementation into clinical practice.”
Collaboration outside the UMCU is also expressly sought with firms that have specific expertise in the field of data analytics. In one such collaboration, we developed an algorithm for the hospital’s neonatology division; it allows us to detect twelve hours earlier than previously that a newborn baby is contracting blood poisoning.
Organisational prerequisites must be fully in place for successful implementation
As data analytics is implemented in daily practice, it is essential to make proper advance arrangements to ensure that all legal, ethical and privacy issues have been clearly regulated. In the UMCU, for instance, arrangements were made concerning the intellectual ownership of the data, and in consultation with our lawyers and the UMCU privacy officer we ensured that the process was airtight from beginning to end. We also reflect on new, still unfathomed questions. How do you determine the reliability of an algorithm? Are doctors legally liable if, on the basis of their expertise and experience, they deviate from the determination made by an algorithm? And what steps are required to validate an algorithm for daily use? A culture shift will be crucially needed in order to get the technological applications successfully embedded into the health care delivery process. More specifically, some health care professionals are embracing the new potentials of technology whilst others are still expressing scepticism. The entire ‘system’, from patients to doctors to managers to external stakeholders, will have to be engaged in the process.
On the eve of an evolution
We will be generating and combining increasing quantities of medical and other data, but danger lurks if we start putting it aimlessly and senselessly to use with no regard to the context of the data. Engaging medical expertise will remain an essential requirement as medical algorithms are developed. Within five to ten years, data analytics will have acquired a permanent place in the daily patient care processes. Data alone, however, leads to ‘mere’ insights. Successful implementation of the techniques and the correct subsequent actions will deliver the added benefits to the patients. That process requires time. Like organisms in nature, we adapt ourselves to a changing environment and we make use of the technological possibilities.
“Step by step, we evolve with the rest, keeping our eye on the goal of improvement and personalisation of care.”
