Developing a clinical decision support system to improve the quality of care for prostate cancer patients

Malignant neoplasm of the prostate (prostate cancer) is the most common cancer among men in the European Union (EU). In fact, about 450,000 men are diagnosed each year. In 2016, prostate cancer was the cause of 3% of all male deaths and 10% of all male cancer-related deaths in the EU, according to Eurostat. In the case of Sweden, it is the number one killer among cancers for men.

Several clinical problems need to be resolved in this type of cancer: curative treatment comprises a trade-off between complete removal of the tumour and the loss of functions relying on tissue close to prostate; tumour location is difficult to visualize, and very radical treatments lead to poor functional outcomes in terms of incontinence and erectile dysfunctions.

Moreover, prostate cancer care involves heterogeneous data from different medical sources (urology, radiology, pathology, oncology, etc.), and data are poorly structured in data bases with very little system integration. In addition to this, the feedback to the care flow is limited and the use of patient-reported data is underdeveloped.

The Prostate Cancer pilot, led by Karolinska within the BigMedilytics project, is developing a clinical decision support system together with Philips to enhance decision-making for surgery in primary prostate cancer patients to be implemented at Karolinska Hospital in Sweden.

The major objective is to enable personalized treatment in order to improve the quality of care for prostate cancer patients and increase productivity. This will be achieved by an integrated visualization of data considering different parameters and a big data-driven predictive modelling. To do this, the partners working in this pilot are gathering in one system and in a structured way the data needed to make the decision, which is a mix of clinical, radiology, and pathology data as well as patient-reported data.

Specifically, the pilot seeks to improve the following aspects:

• Patient outcomes:
  o Better oncological outcome regarding resection margins
  o Better functional outcome concerning urine continence and sexual function
  o Patient satisfaction

• Patient safety:
  o Reduce loss of information in the care flow
  o Better and more efficient decisions at medical treatment conferences

• Efficiency:
  o Register data once, use often (system integration)

• Quality reporting:
  o Real-time quality data, including patient-reported data
  o Stratification by risk group (comorbidity, family history, prognosis, etc.)
  o Real-time, individual feedback to clinical staff