The use of data is crucial to staying healthy for longer. With this in mind, the Board wants to mobilise interested parties to launch an Artificial Intelligence Health Lab in the Amsterdam Metropolitan Area: a gateway for healthcare SMEs and large enterprises, giving access to data, processing capacity and insights. An active coalition is being assembled to make this happen.
One of the participants in the Health Lab is the software company Pacmed. Wouter Kroese explains their approach: “Pacmed makes decision support software that helps doctors to choose between treatment options. This software makes use of large volumes of practical data: data concerning ‘normal’ patients treated under ‘normal’ healthcare. The information about all these patients can help us determine which treatment will work best for a specific patient.
More personal and precise
For example, Pacmed has developed software to support the selection of the most suitable antibiotic. Current knowledge of these drugs is based on mean results over a small group of patients, who often bear little resemblance to the patient before you: the doctor. Based on an analysis of the data from hundreds of thousands of patients, we now see that in older patients with multiple illnesses and likewise medications, relatively uncommon antibiotics are much more effective. Our access to an enormous volume of varied data enables us to see which treatment is best suited to the individual situation of a specific patient. This way we can make healthcare more personal, more precise and more efficient.
Other Pacmed developments include software to support the safe discharge of patients from the intensive care unit, and software that helps psychiatrists choose the best treatment more quickly for patients with severe depression, so that the admission duration of these patients can be shortened by many weeks.
In the development of our software we combine machine learning with medical expertise. Machine learning can find the correlations in the data, but without a medical expert it is impossible to actually arrive at predictive and usable algorithms. Doctors are necessary to understand the data, interpret results, weigh the relevant factors correctly, and to ensure that the results actually have practical value for both doctor and patient. Designing an algorithm is still a hands-on job.
Every step in the development of the software and algorithms is assessed by medical practitioners, to make certain that the data is interpreted correctly. Pacmed has two of its own doctors in house, and we also have a weekly review with doctors from the Intensive Care unit of the VU University Medical Center Amsterdam. We feed that input back into the development process.
The data we use comes from both GP and hospital care. One of the parties working with Pacmed is NIVEL, a national research institute of which 800 GPs are affiliated. They supply us with all structured information from the medical files: the diagnosis, prescription medication and operations submitted to health insurers for payment. The collection of data from hospitals is more difficult. At present we work mainly with academic medical centres that have been collecting data over a longer period, and who also have the facilities to process that data. Our future ambition is to include individual patient experiences in the software, so that we can also determine from their perspective whether, and when, the treatment has had success.
Processes in healthcare are generally slow-paced, but we want to keep moving ahead. In this phase we are mainly addressing the questions for which, based on the currently available data, we can give complete and clear answers, such as a GP’s options for treating a urinary infection, or whether a patient should be discharged from Intensive Care. But of course, what we want most is to use data to make a positive difference for the treatment of as many illnesses as possible.
We are working hard on the necessary IT integration for collection of the data. To enable doctors to spend more time with patients and less on administration, it is important that we move towards user-friendly, interconnected systems. The development and implementation of these systems is a far-reaching process that requires a major investment from all those involved. We are well aware of this, but we believe it would be irresponsible not to make use of all the available data. For this reason, Pacmed is pleased to work with the Board on this initiative. The establishment of a Health Lab contributes to the creation of collective intelligence in the health sector, for the benefit of doctors, patients and society.”
Do you want to stay informed and be connected. Join our network for health innovators on Amsterdamhealth.nl. Sign up here for free.
Interview: Ronne Theunis