Data visualisation using BI for digital health

Abstract

Outpatient services are ever-growing, and several problems are arising when trying to cater for the many patients, thus leading to departmental inefficiencies. Healthcare services have undergone significant changes at an increased rate during the last 20 years. With both patients and clinicians adopting technologies in their processes, the rate of increased change within healthcare systems will continue to hasten. Consequently, the variation in referral rates by General Practitioners (GPs) can exhaust specialist resources. This study thus analyses clinical processes, designs dashboards, and develops an appointment management solution by applying prediction algorithms to generate the likelihood of appointment cancellations. Clinical appointment scheduling processes are designed in BPM Notation to understand better how appointments and referrals are made in the clinic. BI Dashboards are designed to represent such information in an effective format, while facilitating clinical analysis. Furthermore, prediction algorithms are applied on a dataset to generate the likelihood of a patient missing their appointment, hence enabling effective future appointment scheduling. A critical problem within the Urology clinic in the Outpatient is that appointments are usually not scheduled according to criteria, and available slots are often taken up by non-urgent cases. The prediction algorithm enables clinicians to plan for future appointments by knowing which appointments are likely to be missed, thus rescheduling the urgent cases accordingly. Moreover, having dashboards portraying valuable information assists the clinicians in knowing the state of the clinic and patient throughput details. Therefore, meetings were held with the consultant and key clinicians to obtain data and gather requirements. An interview was further scheduled with the clinic consultant to discuss in detail clinical processes, current issues, and KPIs to be visualised in dashboards. A usability study was conducted with five participants to gather their feedback on the dashboards, their functionality, and design. From this study, it was found that the participants within their different roles opted for such dashboards as they found them of assistance to clinicians. The achieved usability score was very high, that is, the dashboards are well designed for this scenario. Since the CPAS system does not currently capture much clinical data, an external dataset had to be sought on which to apply prediction models. The performance of these prediction models yielded very good results, the optimal being the ANN model on unknown data, and thus, it can be considered a solution to appointment allocation in the Outpatient. This study could have been more realistic and applicable if the CPAS system had captured more data on clinics.