Software engineer typing on laptop

    September 9, 2019

    Randers Regional Hospital tests a digital crystal ball


    A new planning and forecasting tool that uses advanced machine learning is able to predict the need for beds and identify risk patients who may need special attention. A digital "crystal ball" of this kind is now being tested at Randers Regional Hospital (Regionshospitalet Randers) in Denmark - and the potential is considerable.

    If hospital staff were able to look into the future and predict how many patients will be admitted to hospital during the week – and to which departments – and how long their stays will be, it would be easier to make sure that there was always a bed available in the right ward. If one could also identify which patients are most likely to experience complications and require re-admission, the hospital staff would then be able to plan the treatment based on this information – and hopefully avoid both.

    Up until now, staff at Randers Regional Hospital organised patient care on the basis of a combination of expectations and experience, but unfortunately this is not always sufficient to avoid overcrowding, inappropriate transfers and re-admissions.

    Big data paves the way for better patient safety 

    But help is on the way. In a forthcoming pilot project, Randers Regional Hospital will test a new planning tool that can predict future admissions with considerable accuracy by using advanced machine learning. The pilot project is part of the Danish national Big Data research project entitled Danish Center for Big Data Analytics driven Innovation (DABAI), and the solution based on this data – bearing the name Columna Patientflow – is being developed by software company Systematic, in collaboration with Central Denmark Region (Region Midtjylland), one of the five regions responsible for health care administration in Denmark.

    The Systematic solution works by feeding a computer with large amounts of historical data covering several years of hospital admissions. The computer then identifies patterns and deviations in the data material, and combines this with current data from the hospital departments to provide solid data on which to base forecasts about future admissions and courses of hospital treatment.

    "This is a scientific project that we are testing in the real world, and if the results are positive, it will be easy to transfer it directly to other hospitals", Anders Goul Nielsen, Group Senior Vice President for Healthcare in Systematic.

    Digital crystal ball helps doctors make key decisions

    The Systematic solution can also identify patients likely to be re-admitted or to experience complications during their stay in hospital. This makes it possible for the Columna Patientflow system to provide data to support the decisions that doctors and nurses need to make when planning the course of treatment for a patient at risk.

    One of those looking forward to trying the system is Bjarke Johannesen Bruun, consultant and specialist physician in the Internal Medicine department at Randers Regional Hospital.

    - It will be a great help if the system can alert us about patients who are at risk of complications and re-admission, so that we can plan the course of treatment to take this into account and make sure the patient receives the appropriate support in the home afterwards. The system cannot replace medical assessments, but it will be a great tool to help us in our work and perhaps make it easier to identify possible inefficiencies or inconveniences in patients’ courses of treatment, he says.

    Bjarke Johannesen Bruun also highlights the possibility of achieving better dialogue with patients as an obvious benefit of the system.

    - Sometimes a patient wants to go home quickly, and it’s not always easy to hold on to people if they feel healthy enough to be discharged. But if the medical prognosis and our professional judgment both point to an increased risk of re-admission, we have a good background for a fact-based dialogue with the patient about this, he says.

    Good experience with smaller solution in northern Denmark

    Randers Regional Hospital is the first place in Denmark to use such an advanced solution. Good experience is currently being harvested from another somewhat similar model in use at six hospitals in the North Denmark Region (Region Nordjylland). Here, 56 hospital departments receive daily machine learning-based projections for bed occupancy in the coming week, together with an overview of the actual number of available beds in the wards, in real time. This transparency provides far better opportunities for organising safe, effective patient treatment, while also reducing waiting times and transfers.

    The patient flow project in Randers uses the same model, but also has built-in forecasts based on historical data about patients’ progress. This makes it possible to predict the length of the courses of treatment and the likelihood of re-admissions. The project will first be taken into use at Randers Regional Hospital, after which the plan is to roll the system out to other hospitals in the region. It is expected that the staff at the Randers hospital will be able to make use of the system during the autumn of 2019, and it will be tested until February 2020.

    - This is a scientific project that we are testing in the real world, and if the results are positive, it will be easy to transfer it directly to other hospitals, explains Anders Goul Nielsen, Group Senior Vice President for Healthcare at Systematic.

    Det forventes, at personalet i Randers kan tage systemet i brug i løbet af efteråret, hvor det skal testes frem til februar. 


    More information

    Central Denmark Region (Region Midtjylland): Quality and Health IT Manager, Randers Regional Hospital
    Thomas S. Pinstrup, +45 6168 0430, [email protected] 

    Systematic: Director Business Development & Product Management, Healthcare Operations
    Mikkel Harbo, +45 2544 2803, [email protected]