Selected papers from Koa Health

We’re dedicated to sharing our knowledge with the larger scientific community and leveraging the benefits of peer review to create better health solutions for people everywhere.

That’s why we openly communicate our findings in industry publications such as Nature Medicine, JMIR MHealth Uhealth, ACM Press, and Behavior Therapy.

For further details, access our full list of papers to date, or keep scrolling for a selection from our research team.

Testing the use of a mental health and wellbeing app in a healthcare worker population

The largest known study of its kind researching the use of a mental health and wellbeing app in a healthcare worker population has been published by the Institute of Psychiatry, Psychology & Neuroscience (IoPPN) at King's College London. This study reveals that the mental health and wellbeing of frontline NHS healthcare workers can be improved through the use of Koa Foundations.

S. Gnanapragasam, S. Nishanth, R. Tinch-Taylor, H. R. Scott, S. Hegarty, E. S. Souliou, R. Bhundia, D. Lamb, D. Weston, N. Greenberg, I. Madan, S. Stevelink, R. Raine, B. Carter and S. Wessely. A multi-centre, England wide randomised controlled trial of the "Foundations" smartphone application in improving mental health and wellbeing in a healthcare worker population. British Journal of Psychiatry (2022): 1-18.

Testing the efficacy of App-based CBT for Body Dysmorphic Disorder with Coach Support

Body dysmorphic disorder (BDD) is severe, chronic, and undertreated. Self-help options, particularly app-based solutions, could substantially improve treatment and outcomes for BDD. Our recent research tests the usability and efficacy of Perspectives, our coach-supported, app-based cognitive-behavioural therapy (CBT) solution for BDD.

S. Wilhelm, H. Weingarden, J.L. Greenberg, S.S. Hoeppner, I. Snorrason, E.E. Bernstein, T. H. McCoy, O. T. Harrison. Efficacy of App-Based Cognitive Behavioral Therapy for Body Dysmorphic Disorder with Coach Support: Initial Randomized Controlled Clinical Trial. Psychotherapy and Psychosomatics. 1-9.

Machine learning algorithm to predict mental health crises

Experiencing a mental health crisis has a detrimental impact on a patient’s life. Can science help predict and prevent mental health problems? A machine learning model created by our R&D team based on 7 years of electronic health record data from 17,122 patients has shown to help clinicians predict mental health crises.

R. Garriga, J. Mas, S. Abraha, J. Nolan, O. Harrison, G. Tadros, A. Matic, Machine learning model to predict mental health crises from electronic health records. Nature Medicine, 2022.

CBT via smartphone

Smartphone-delivered cognitive behavioural therapy (CBT) has become more common over time, but it still doesn’t use smartphone capabilities to their full potential. Can we leverage these capabilities more effectively to optimise treatment?

"Optimizing smartphone cognitive behavioral therapy for body dysmorphic disorder using passive smartphone data: Initial insights from an open pilot trial", JMIR MHealth Uhealth, 2020.

Development and testing of CBT for BDD

Cognitive behavioural therapy (CBT) has been shown to be effective in the treatment of body dysmorphic disorder (BDD) but access is an issue. Could user-centred treatment via smartphone provide a feasible, acceptable and satisfactory solution to the problem of accessibility?

"Development and pilot testing of a cognitive behavioral therapy digital service for body dysmorphic disorder", Behavior Therapy, August, 2019.

Mood prediction models in bipolar disorder

Sensors already embedded in our phones have the potential to anticipate moods swings in bipolar disorder to enable more timely treatment. Can these prediction models for this purpose be generic, or do they need to be personalised to each patient?

"Personalized versus generic mood prediction models in bipolar disorder", Mental Health and Well-being: Sensing and Intervention Workshop in conjunction with UBICOMP 2018 Conference, Singapore, 2018.

Technology and mental health: Insights from a decade of research

A significant body of research in ubiquitous computing is devoted to the design of technologies for the continuous monitoring, diagnosis and care of mental health conditions. By analysing the last ten years of research in this field can we better address the technical and clinical challenges of designing ubiquitous computing technology for the decade to come?

"A decade of ubiquitous computing research in mental health", IEEE Pervasive Computing 19, no. 1, 2020.

Expert views on the future of pervasive health

Pervasive health is about bringing people and knowledge together to create higher performing healthcare. But what must be done to create pervasive health solutions? What technological challenges, adoption and adherence barriers, and ethical issues must be addressed?

"The Future of pervasive health", IEEE Pervasive Computing Journal, vol. 16, no., pp. 16-20, 2017.

Mobile network data for mental health

Smartphones generate an unprecedented amount of human behavioural data at individual and aggregated levels. But how can we apply this data to public health? What are the opportunities and challenges involved?

"Mobile network data for public health: Opportunities and challenges", Frontiers of Public Health, 2015.

Aligning daily activities with personality

When it comes to improving health and subjective wellbeing, recommender systems haven’t been sufficiently explored. And while recent advances in smartphone-based technologies and user modelling present opportunities for delivering such systems, understanding the drivers of subjective wellbeing at an individual level is key. So the question becomes, how can the congruence between activities and personality traits be maximised to improve subjective wellbeing?

"Aligning daily activities with personality: Towards a recommender system for improving wellbeing", Proceedings of ACM Conference on Recommender Systems (RecSys), 2019.

Cultural impact and automatic personality models

To make greater technology personalisation possible, sensor data collected from smartphones is used to infer users’ personality traits. But improving the accuracy of this type of inference is a significant challenge, especially when many models don’t consider the potential impact of culture on the data obtained. Could country-specific datasets improve trait prediction? And if so, by how much?

"Modeling personality vs. modeling personalidad: In-the-wild mobile data analysis in five countries suggests cultural impact on personality models", Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies archive (IMWUT), Volume 3 Issue 3, Presented at Ubicomp, London, UK, September 2019.

Data minimisation and personality classification

Passive personality modeling can help personalise apps and phone-based services but generally requires weeks or months of data collection and can negatively impact user engagement. But what if a few weekends would suffice for accurately classifying personality traits towards improved user engagement?

"Personality is revealed during weekends: Towards data minimisation for smartphone based personality classification", INTERACT conference, 2019.

Data minimisation and color blindness

Data minimisation is a useful tool in wellbeing recommender systems. But if it’s applied too generally across the board, can it lead to problems with color blindness? If so, what are the implications of this color blindness?

"Auditing algorithms: On lessons learned and the risks of data minimization", ACM Press, 2020 .

Explainability in AI

Explainability in Artificial Intelligence is a crucial part of conveying safety and trust to users in the "how" and "why" of automated decision-making. TREPAN, an algorithm that explains artificial neural networks by means of decision trees, can be extended and used to generate more understandable explanations. But are these explanations more understandable for users than those generated by standard (unmodified) TREPAN?

"Trepan reloaded: A knowledge-driven approach to explaining black-box models", Proceedings of the 24th European Conference on Artificial Intelligence (ECAI), 2020.

Building a system to predict empathy

Researchers describe a system based on deep-learning for empathic emotion recognition developed to process multiple data streams, aka modalities integrated at a late stage (late fusion). Could this modular approach allow for ease of development, optimisation and future iterations?

"Towards a multimodal time-based empathy prediction system", 14th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2019) IEEE, Chicago, 2019.

Explaining intelligent machines

What makes a good explanation, really? By exploring the cognitive dimensions of explaining intelligent machines can we identify what makes this process successful (and understandable) or not?

"What makes a good explanation? Cognitive dimensions of explaining intelligent machines",Proceedings 41th Annual Meeting of the Cognitive Science Society, CogSci 2019, Montreal, Canada, pp. 25–26, 2019.

On the comparator model and sense of agency

A sense of agency is an indispensable part of how cognitive entities (biological and artificial) become cognitive agents. Can the comparator model used in developmental psychology to assess agency in early infancy be effectively used in robotics and artificial systems research?

"A match does not make a sense: on the sufficiency of the comparator model for explaining the sense of agency", Neuroscience of Consciousness, vol. 1(niz006), 2019, Oxford Academic/Oxford University Press, 2019.