It is important to provide patients with reliable, accurate and personalised information about their health conditions, to engage them in their health. Often, this is done by doctors, based on their textbook knowledge and prior clinical experience in communicating with and managing similar patients under their direct care. Whilst this usually works well for experienced doctors and receptive patients, not all doctors have the same wealth of experience and it can be hard to advise patients of the need to accept life-changing therapies or to adopt lifestyle changes. Consequently, we propose to develop an informatics dashboard to augment patient-doctor decision-making. This would quantify and demonstrate disease-related risk for an individual compared to other patients who have the same disease but variations in other health factors.

Doctors have mental models of their patients, populated from health records and from face-to-face consultations. These mental models include the patient’s demographics, past medical history, family history and current state of health. Each doctor has to use his or her clinical knowledge and experience to determine the ‘weight’ or importance of an aspect of the mental model in the disease being considered. For instance, smoking is important as a risk factor for cancer, but it is unlikely to be important as a risk factor for traumatic injury.Precision Driven Health is supporting this research project which aims to generate research and technology that assists doctor-patient decision making. First, we aim to develop and evaluate an algorithm-based approach for determining the ‘similarity’ between patients, based on various user-specified attributes, captured in the patient’s electronic health records. This patient similarity measure can then be used for finding ‘similar’ patients in a population. We would work with doctors to define the attributes to select based on what weight they would assign to them.

Second, we aim to develop an interactive dashboard that will not only display the “progression of disease” trajectory of similar patients (over a certain timeframe), but also provide the ability to make adjustments to the progression trajectory in real-time by updating different variables such as treatment plans, lifestyle choices and medications. This would augment rather than replace doctors’ discussions, for example, where statistical inference is unreliable due to low disease frequency, novel therapies, etc. We believe patient care can be improved through using data science to understand decisions that other doctors have taken with similar patients, building on an individual clinician’s knowledge by supplementing it with the collective experiences of others who have treated similar patients.

As part of this project, the research team is conducting interviews to learn about shared decision making in cancer care. Find out more!