Artificial Intelligence to enhance cancer care

Digital & Innovation: The Centre for Data Analytics and Cognition (CDAC) at La Trobe University has teamed up with international cancer researchers to create the Patient-Reported Information Multidimensional Framework (PRIME) for detection and analysis of patient behaviours, clinical factors, decision-making and deep emotions when dealing with cancer.

Using new Artificial Intelligence (AI) algorithms and natural language processing (NLP) techniques to make sense of vast quantities of information, Professor Damminda Alahakoon – Director of CDAC and Analytics Discipline Head at La Trobe University – said PRIME can better understand a patient’s mental health status based entirely on the data shared by the patient.

“This data can be the text a patient provides to an online chatbot, an online cancer support group or other online support service,” Professor Alahakoon said.

“Our first study – published in PLOS One – analysed 609,960 conversations from 22,233 patients, which comprised of 93,606,581 words.

“PRIME addresses the challenges associated with understanding the unlabelled and unstructured nature of this data, allowing it to efficiently identify trends and anomalies – such as when a patient is struggling emotionally – and effectively adapt to the changing nature of that data.”

He said PRIME provides much deeper insights into patient emotions, needs and expectations during their cancer journey.

“Our technology can provide health professionals with the support they need to better monitor their patients,” Professor Alahakoon said.

“PRIME can be that crucial first-step in identifying a patient’s mental health needs and alerting health professionals to those who are at high risk if untreated.”

PRIME was developed in collaboration with a team of oncology urologists treating prostate cancer patients at the Austin Hospital in Melbourne and later with psycho-oncology researchers from Canada.

Austin oncologist Professor Damien Bolton said the team had noted a lack of insight into emotional distress meant most clinical decisions are based on research trials, rather than individual patient needs.

“Men use online support group environments to address their unmet needs, express emotions and voice their worries freely,” Professor Bolton said.

“PRIME can help analyse the outcomes of those patient-reported information.”

PRIME was further developed after a team of psycho-oncology researchers at the University of Toronto and De Souza Institute in Canada asked La Trobe to design and develop an Artificial Intelligence-based Co-Facilitator (AICF).

Assistant Professor Yvonne Leung, from the Department of Psychiatry at the University of Toronto said AICF is designed to track and monitor patient emotions in real-time during professional-led online cancer support group sessions.

“AICF, our adaptation of PRIME, was able to detect ‘patient distress’ with a baseline accuracy of 78 percent. But its true value lies beyond accuracy, as it dissects distress into constituent emotions of fear, sadness, anxiety and anger,” Assistant Professor Leung said.

Professor Alahakoon said La Trobe’s promising work with Austin Hospital and the University of Toronto could result in the expansion of PRIME across other medical and healthcare areas in the world.

The research findings from these collaborations have been published in PLOS One, Annals of Surgical Oncology, BJU International, Urologic Oncology: Seminars and Original Investigations, and the Journal of the Association for Information Science and Technology (JASIST).