Biased Artificial Intelligence: Openness, innovation and the remaking of mental health

Informazioni relative al progetto

Biased AI

ID dell’accordo di sovvenzione: 794604

DOI

10.3030/794604

Progetto chiuso

Data della firma CE 9 Aprile 2018

Data di avvio 1 Settembre 2018

Data di completamento 31 Agosto 2020

Finanziato da

• EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions

Costo totale

€ 172 800,00

Contributo UE

€ 172 800,00

Coordinato da

UNIVERSITEIT GENT
Belgium

Periodic Reporting for period 1 - Biased AI (Biased Artificial Intelligence:Openness, innovation and the remaking of mental health)

Periodo di rendicontazione: 2018-09-01 al 2020-08-31

Sintesi del contesto e degli obiettivi generali del progetto

The particular problem to be addressed concerns the possibility of powerful entities using AI in ways that fail to take into account the health needs of patients. Governments may use AI to organise more efficiently government itself and private entities may want to profit from their monopolistic position.

Importance for society: Ultimately the project will seek to contribute to the debate on designing ‘good AI’ by means of making concrete policy proposals in favour of a system that does not lose sight of the individual and as such promotes responsible innovation in the field.

Objectives: AI is a tool to mine troves of health data. Therefore, the first objective consisted of identifying novel business models and partnerships between the private and public sector in the emerging health data market. The second objective concerned the role of intellectual property rights and 'open' models of innovation and what they tell us about the way AI is currently being developed. The third objective concerned discussing public policy implications of using AI in wearables and mobile apps ito nudge people to change harmful behaviors such as eating habits and non adherence to medication and link this discussion to the problem of responsible innovation.

To meet these objectives qualitative research and analysis of legal/policy documents was conducted to identify public and commercial demands in order to rethink the governance of innovation in AII. Participant observation and Interviews with key stakeholders were conducted to elicit their views. The governance question needs to take into account public demands that are technically feasible and at the same time commercially viable in the AI era.

Lavoro eseguito dall’inizio del progetto fino alla fine del periodo coperto dalla relazione e principali risultati finora ottenuti

September 2018-February 2019: literature review on uses of AI in medicine; privacy and the legal framework regulating the uses of Software as a medical device in the US and Europe; algorithmic fairness and accountability. I applied this new knowledge to map the actors involved in the development of AI (hospitals, private entities, governments) in the context of a case study on digital pills and uses of AI for monitoring medication intake.

September 2018: attended (with Prof. Sigrid Sterckx) conference 'AI and Law' at the University of Cambridge.

15-16 November 2018: attended (with Prof. Sigrid Sterckx) the launch event (workshop) of the Centre for the Study of Contemporary Solidarity of the University of Vienna, with a strong focus on Big Data in Biomedicine.

December 2018: Conference presentation "Artificial Intelligence, Personalised Medicine and Intellectual Property Rights: Openness Revisited" at the University of Antwerp.

May 2019: job interview at Panteion University, Department of Political Science and History, Athens Greece. Job offer made and it was accepted. Job started on 1 September 2020.

June 2019: paper presentation "Bias in AI and digital phenotyping" conference, Radboud University Nijmegen, The Netherlands.

September - October 2019: Secondment with Prof Tsmardinos' AI startup in Heraklion, Crete. Professor Tsamardinos has developed highly innovative machine learning techniques to mine health data.

September 2019: attended the MYDATA 2019 conference in Helsinki, Finland.

October 2019 - February 2020: three research visits to allow dissemination and networking: Vrije Universiteit Brussel in Brussels, University of Oxford, and Kings College London. I gave a talk at Kings College and the VUB. I taught a seminar to Postgraduate students at the University of Oxford.

October 2019: Lecture at Ghent University, Faculty of Medicine, by invited speaker Professor John Harris, "Automated algorithm-based decision-making: Ethical and legal aspects" (organised by Prof. Sterckx).

November 2019 - May 2020: 25 interviews with stakeholders. Interviews await analysis. Also dissemination to postgraduate students at University of Crete (5 seminars).

December 2019: publication ‘Remote digital monitoring of medication intake: Methodological, medical, ethical and legal reflection’ in the peer reviewed journal Acta Clinica Belgica (with co-authors Prof. Van Biesen, Decruyenaere, and Sterckx, and Dr Cockbain)

January 2020 : paper submitted to peer reviewed journal Ethics and Information Technology, 'Digital pills and ‘Artificial Intelligence’ image analysis: Ethical and legal reflections on competing interests in the market for remote monitoring of medication intake’ ' with co-authors Prof Sterckx, Van Biesen, Decruyenaere, and De Hert, and Dr Cockbain. (paper still under review – review delayed due to COVID-19)

March 2020: publication: Sideri, K. “Openness and the Political Economy of Intellectual Property Rights.” Science and Public Policy.

April 2020: Publication: K. Sideri & G. Dutfield (eds.), Openness, Intellectual Property and Science Policy in the Age of Data Driven Medicine, Special Issue of Science and Public Policy

April 2020: publication: G. Dutfield and K.Sideri editorial introduction to the aforementioned special issue.

May 2020: research proposal submitted to the Greek fund for Higher Education (ELIDEK) with the title 'Smart Therapeutics'.

June 2020: acceptance of a panel proposal with colleagues from the Vrije Universiteit Brussel for a conference at the University of Twente (to take place fall 2020 (title of the paper: Agonistic algorithms and Medical software; this paper will be based on the analysis of interview data (see above).

August 2020: paper submitted to Journal of European Public Policy. Title: Persuasion in a sensor as a choice of policy: Data driven nudges, structural bias and health policy.

August 2020: Paper submitted to Research Policy (with Andreas Panagopoulos Timo Minssen and Helen Wu (University of Copenhagen). Title 'Valorizing and Governing Healthcare Data in the AI Era.'

Spring 2021: In December 2019 we agreed to organize a workshop at Kings College London. This has been postponed due to COVID-19. If it takes place next year, the Marie Curie programme will be acknowledged.

Progressi oltre lo stato dell’arte e potenziale impatto previsto (incluso l’impatto socioeconomico e le implicazioni sociali più ampie del progetto fino ad ora)

The project went beyond the state of the art since:

1. Publications addressed a much neglected area, the political economy of intellectual property rights and how this perspective affects our understanding of how AI is currently being developed.

2. The research outputs include a novel understanding of how innovation in the field is driven by specific actors and their needs. I conducted a series of Interviews with stakeholders.

3. Research outputs (in the form of an article submitted to Research Policy) include a completely novel proposal to use the management structure of a collecting society for health data and AI.

4. Research outputs (in the form of an article published in Acta Clinica Belgica and a second article submitted to Ethics and Information Technology) offered an insightful analysis of a variety of different ethical and regulatory issues concerning the uses of of AI to monitor medication intake.

Generally, the analysis went beyond what was envisioned in the initial proposal (a more narrow account of how Intellectual property rights interact with open models of collaboration) as a more nuanced picture emerged that required addressing novel business partnerships along with the wider question of collaborative models of governance. We believe that in this way the project is more attuned to the needs of policy makers to understand the wider issues surrounding the uses of AI in medicine.

image2.jpg