Make the most out of big data in service planning

When big data technology being discussed across sectors, since early 1990s Hospital Authority (HA) has indeed established by phases a Clinical Management System to cope with clinical and operational needs, and has started to analyse structured clinical data in the System to facilitate planning and formulation of various healthcare services.

Community Health Call Centre service has leveraged on an automated tool to identify elderly patients with a high risk of re-admission to hospital.

For example, around 10 years ago, in response to the need to formulate the Hospital Admission Risk Reduction Program for Elderly (HARRPE), HA has already analysed a huge volume of clinical data and developed a risk prediction model according to 14 predictor variables. Supported by information technology, this tool has been automated to calculate the predicted chance of hospital re‑admission (known as HARRPE score) for elderly patients from medical wards every day. Currently, Community Health Call Centre and Integrated Discharge Support Programme for Elderly Patients are making use of this tool to screen elderly with a higher risk of re‑admission, and enhance their post-discharge support in order to reduce their chance of re‑admission.

Dr Cheung Ngai-tseung, Head of Information Technology and Health Informatics (left) and Eva Tsui, Chief Manager (Statistics & Workforce Planning) both opine that the shortage of talent is the biggest challenge. To synergise with the clinical teams, HA needs talent who are familiar with information technology and analytics techniques, and understand the healthcare system.

Recently, HA has started to explore the use of unstructured clinical data such as radiological images and clinical reports for data analytics. Eva Tsui, Chief Manager (Statistics & Workforce Planning) says, “We are only at a beginning stage to explore how to make use of these unstructured clinical data. We are now working with a local university to undertake a research study related to acute stroke. In the first phase, we provide various data of relevant cases, after removing individually identifiable information, to the neurosurgeons and statistical experts in the university, who will analyse the brain images and other clinical data, such as age, medical history and major clinical symptoms, in an attempt to develop an automated real-time risk prediction model. In the second phase, the model will be applied to more cases to test its feasibility of applying to clinical systems and services.”

However, we have to strike a balance between data disclosure and personal data privacy protection when providing clinical data to third parties. HA has made reference to the practices in some developed countries such as the United Kingdom and Australia, and all along adhered to the ‘Five Safes’ (5S) principle.

Safe Data: Reduce the possibility of re-identification of individuals from the data. Provision of data or information leading to patient re-identification by direct or indirect means should be avoided.

Safe Projects: Use of data in the projects is lawful and with moral and ethical considerations.

Safe People: Researchers possess adequate knowledge and skills to use and store data appropriately.

Safe Settings: The facilities and methods for data access ensure adequate data protection.

Safe Outputs: Research or analysis results will not disclose the identity of individual patients.

Good use of data for accurate and insightful analytics helps formulate and better plan for healthcare services. Nevertheless, the key consideration is to ensure data security and personal privacy protection.

Eva adds that all along HA has an established mechanism to process and vet external data requests (mostly from universities). Based on the 5S principle, ‘de‑identified’ data would be provided to external parties for academic research, with a view to facilitating local healthcare research. In addition, a wide range of statistics is also available for public use and can be downloaded from the HA corporate website and government’s public sector information portal (https://data.gov.hk/en).