Measuring outcomes from the patient’s perspective represent one of the fundamental principles in cancer survivorship. Value in healthcare should always be defined around each induvial and outcomes rather than volume of services delivered. Patient-reported outcomes (PROs) represent real-world data, in the form of self-reports from individuals and their experiences. PROs collected in routine daily life can improve decision making, by providing insight into quality of life, such as the ability to maintain one’s roles and responsibilities, and in proactive symptom management. However, there are several challenges regarding efficient collection of the patient-reported outcomes and their integration into the clinical workflow. Patient adherence, data linkage and scaling across health care systems (interoperability), provider engagement, and actionability are recognized as main challenges. Although advancements in technology have the potential to revolutionize medicine through the collection and analysis the following points must be considered:
- How to efficiently collect data from patients?
- The cost and time for collecting PROs?
- How to integrate data into clinical workflow?
- How to enable proper interpretation by the clinicians?
Embodied Conversational Agents To Assist Survivors
The paradigm of value-based healthcare represents an important shift towards more efficient and more effective medical care. Conversational intelligence and Natural Language Processing, i.e. Chatbots, are one of the main digital technologies that can significantly contribute to patient activation and engagement. These artificial entities are cost-effective and available 24/7, and can support patients even outside their doctor’s operating hours. However, chatbots are yet to tackle long-term adherence with a sustainable quality of the reported data. Embodied conversational agents (ECAs) extended the traditional chatbots with a virtual body and offer a full human-like experience. ECAs can engage with users in more diverse interactions enriched with empathy, gestures and facial expressions. In fact, ECAs can deliver a system with symmetric multimodality, having speech, gesture, facial expression on both the input (survivor) and the output side (virtual agent). The fully symmetric interaction opens up the opportunity to introduce human-like qualities, significantly improving the believability of the interfaces and sustainability of the digital interventions. Namely, sharing of information in human social interactions is far more complex than solely an exchange of words. In interpersonal discourse, the verbal signals carry a symbolic or semantic interpretation of information, while gestures and facial expressions orchestrate speech and allow individuals to express moods, attitudes and social functioning. Understanding these signals and even recrating them makes ECAs a ‘safer’ interaction partner that never judges.
In PERSIST, we deliver a multimodal network of digital services delivering digital interventions to collect real-world data. The network consists of:
- a multilingual virtual assistant; a fully articulated ECA implementing symmetric interaction in 6 languages and with male and female representation
- a microservice-based sensing network to support spoken language and collect patient information through PROs and by understanding the conversation between individual and the agent
- a holistic approach towards interoperable and fully integrated real-world data
University of Maribor’s (UM) Role in PERSIST
UM’s role in PERSIST is to empower Survivors and Clinicians with the huge potential hidden in the real world data. To this end UM leads WP4, dedicated to determine and monitor the combined effects of: (i) cancer treatment, environment, (ii) lifestyle and wellbeing parameters, and (iii) genetics, on the quality of life of cancer survivors. UM delivers the patient sensing network, involving soft- and physical sensors, as new sources of health data originating mobile health apps, wearables, conversational agents and chatbots and, innovative biomarkers methodologies, and supported by multilingual, multimodal embodied conversational agents. The Virtual Assistants support all PERSIST languages, English, French, Latvian, Slovenian, Spanish and Russian.
Author’s name and position:
- Dr. Izidor Mlakar, principal researcher in project PERSIST, University of Maribor, Faculty of Electrical Engineering and Computer Science, Laboratory for Digital Signal Processing
- Assoc. prof. dr. Bojan Musil, head of the department of Psychology, University of Maribor, Faculty of Arts, Department of Psychology