Type:
Educational Exhibit
Keywords:
Artificial Intelligence, eHealth, Neural networks, eLearning, Technology assessment, Cancer, Multidisciplinary cancer care, Outcomes
Authors:
K. Ballurkar1, Y. Chshipunova2, J. Yee2, J. Lo1, Z. Tan1, R. Murali-Ganesh1; 1Eveleigh, NSW/AU, 2Sydney, NSW/AU
DOI:
10.26044/ranzcr2021/R-0324
Background
Our organisation runs an outpatient commercial digital integrative cancer care service that targets working-age patients who have discontinued work after cancer diagnosis and treatment. The service is multifaceted, is digitally advanced and encourages overall wellbeing and return to regular life. Participants can enrol in the service through a variety of means and partners; in this study we retrospectively analysed 365 cancer patients that commenced the service, who paused work and claimed income protection through a single life insurer on account of the burden of their cancer and treatment. See Figure 1. Another study currently in pre-print has shown this cancer care service significantly increased the rate of return to work in cancer patients compared to a propensity-matched equivalent cohort.
The most common cancers were breast, colorectal or genitourinary in origin, however there was a broad variety. 40% of our patients reported receiving radiotherapy. The program runs over several weeks and combines human and digital touch points to hone wellness in a variety of domains such as exercise, diet, mental health and return to work strategies. See Figure 2.
Patients receive regular telephone calls from trained health coaches (medical doctors or nurses) to assess their cancer recovery, for which the coaches document clinical progress notes. Patients also engage with a range of digital multimedia such as online modules, videos, articles, and symptom tracking, and they can be referred to other practitioners (e.g. exercise physiologists and counsellors).
The digital service should be tailored to the patients’ specific requirements. Currently this is a time-intensive task based on manual review of coaching progress notes. For example, notes that discuss physical ailments such as pain, fatigue or dermatitis would prompt prioritisation of in-app symptom tracking, articles to help with fatigue, encouragement to liase with their GP or appropriate allied health referral. For patients with psychosocial stressors, in-app mood tracking and mental health videos and articles can be offered. Some patients have both poor physical and social wellbeing.
Clinical progress notes are abound with information but this is unstructured and difficult to harbour insights from. Natural language processing (NLP) is a domain of artificial intelligence (AI) that is commonly deployed to process and analyse human-written text data. One of its many capabilities is to classify text data into useful categories. In collaboration with computer science students from the University of New South Wales, we piloted NLP to retrospectively analyse our clinical progress notes. Patients were categorised as requiring more support or not in each of two domains - physical and social wellbeing.
NLP is already used in radiology and radiation oncology - including quality assessment, cohort building and diagnostic surveillance.[1] It has an important future in analysing the vast volume of free-text clinical data that is generated. In our example, we applied a relatively simple NLP model to retrospectively evaluate our notes to identify increased care requirements. Ultimately this can optimise service provision and help catalogue our patients' needs.