Is NOT coming… it IS HERE already in many places across health and care NHS Scotland: examples of AI Projects Numerous Artificial Intelligence projects are underway in the NHS in Scotland. AI-Powered Chatbot NHS Inform AI-Powered ID Verification for Covid status app Imaging (Radiology) Diagnosis Robot assisted surgery Health monitoring Predictive health analytics NLP (Natural Language Processing) applied to unstructured health data Admin tasks Exploring use-cases in Scotland for drug discovery and development and personalised treatment, as well as many other application Did you know these days some vehicles are AI-powered? We also used them across health and care services… Source: https://www.radiology.scot.nhs.uk/projects/artificial-intelligence/ • Privacy and security • Bias in Algorithms • Used with lack of regulation • Interoperability challenges • Clinical adoption and trust • Job displacement • Black boxes • Early Diagnosis and Predictive Analytics • Personalized Medicine • Efficient Data Management • Robot-Assisted Surgery • Drug Discovery and Development • Remote patient monitoring • Better medical records (e.g. transcription of patient-clinical interactions - NLP) • Virtual health assistance • Chatbots for mental health • Administrative efficiency • Social care and elderly support Why we care? Why SIROs should care about AI? • AI introduces a new set of risks and challenges from the information security point of view • Confidentiality • Integrity (e.g. data quality, accuracy etc.) • Availability of data and information systems • AI systems often deal with large amounts of sensitive data, and if not properly secured, they can become vulnerable to cyber threats or lead to inappropriate decisions • Compliance (e.g. data protection, medical devices, NIS, BSI 30440 AI in healthcare etc.) • Ethical considerations: moral consideration and risk of public backlash • Dependency on AI - far reaching consequences, resilience and continuity of health & social care services What are we doing about AI? Working closely with colleagues to identify opportunities, risks and develop guidance and resources • Recognition of AI potential • Assessing AI impact • Availability to the public at a very fast pace with little to none guidance and understanding of risks • Providing direction and guidance AI Policy development (Strategic direction and guidance) AI Regulation (Analysis, advise, legislation, promoting standards) Stakeholder engagement (advocacy groups, AI industry experts, innovators, etc) AI Research and Analysis Analyse evidence, evaluate best practices, and assess the potential impact of AI and AI-related policies on different demographic groups or sectors AI-related assurance and compliance (frameworks, guidance and monitoring) AI Direction and guidance • Collaborative work is ongoing (national, UK-wide, international). Contributing to shape AI standards and regulations • AI Task force • • • • • Scotland’s AI Strategy (2021) EU AI Act ICO AI and data protection guidance / toolkit BSI 30440 & Alan Turin Institute AI Code of Conduct/Practice …more standards coming • Linking and aligning with the wider AI work e.g.: • Scottish AI strategy and register • Scottish AI Alliance (scottishai.com) • West of Scotlland Innovation Hub (exploring AI opportunities, e.g. AI skin cancer consortium project – FV) • Supporting current AI innovations in health care • Advice / Guidance • AI Regulations advice service • Developing resources • e.g. AI Chatbot applications – baseline IG pack • Linking with wider resources: • scottishaiplaybook.com • scottishairegister.com Guidance on the use of Generative AI in the Civil Service Our commitment • Digital Health and Care in Scotland is committed to the ethical, transparent consideration of use of AI based tools. • Any tailored AI policy will reflect this commitment. Scottish AI Alliance: The Data Lab and the Scottish Government Examples of application in Local Government • Traffic management • Public safety (e.g. crime patterns) • Chatbots for citizen services • Predictive maintenance (bridges, roads, etc.) • Data analytics for decision-making • Waste management (eg. collection routes, reducing unnecessary pickups in conjunction with using smart bins etc.) • Citizen feedback analysis • Fraud detection (e.g. ensuring that benefits reach those who genuinely need them). • Emergency response optimisation • ….and many more. LAs generate vast amounts of data. AI can help making sense of this data, enabling informed decisionmaking on issues like urban planning, resource allocation and policy development. Examples of application in Social Care • Personalised Care Plans AI is increasingly being integrated into social care services. • Remote monitoring e.g. elderly or vulnerable populations, allowing them to stay at home while receiving proper care. It is becoming a valuable tool in enhancing social care services, making them more efficient, personalised, and responsive to the diverse needs of individuals and communities. • Support Chatbots • Predictive analytics for health and wellbeing interventions • Social worker support (e.g. caseload management, identification of high-risk cases, streamline paperwork, suggest interventions etc.) • Speech and emotions recognitions – detect signs of distress or mental health issues (e.g. in conjunction with telehealth & telecare services) • Caregiver assistance (guidance, reminders, etc.) • Robotics in care facilities (companionship, monitor vital signs, assisting with daily living tasks) • Child protection: e.g. to identify pattens that may indicate potential cases of abuse or neglect. • Language processing for social services applications: eg. to streamline the application process to access services. This Photo by Unknown Author is licensed under CC BY BS 30440:2023 Validation framework for the use of artificial intelligence (AI) within healthcare. Specification. Single comprehensive resource Concretely actionable and auditable Third Party auditing Using BS 30440 can: • Enable the validation of AI in healthcare • Accelerate innovation in AI healthcare systems • Facilitate trade in AI healthcare solutions • Improve the efficiency with which beneficial solutions are developed and deployed • Increase the confidence in adoption of AI in healthcare • Strengthen the risk management of both suppliers and purchasers of AI systems used in healthcare setting Introducing BS 30440 Validation framework for the use of AI in healthcare (youtube.com) BS 30440:2023 Validation framework for the use of artificial intelligence (AI) within healthcare. Specification. What does BS 30440 - Validation framework for the use of AI do? Why should you use BS 30440 - Validation framework for the use of AI? Gives a comprehensive set of requirements for key evaluation criteria such as clinical benefits, standards of performance, successful and safe integration into the clinical work environment, ethical considerations, and socially equitable outcomes from system use Is both a guide for the development of AI systems for use in healthcare, and a means to assess them for conformity and certification. Introducing BS 30440 Validation framework for the use of AI in healthcare (youtube.com) It will help suppliers develop products that demonstrate efficacy and reach sufficient standards of technical performance. Healthcare providers, clinicians and patients can gain assurance that AI systems will integrate safely into the clinical working environment, are clinically effective and ethical. PREVIEW: BS 30440:2023 | 31 Jul 2023 | BSI Knowledge (bsigroup.com) BS 30440:2023 Validation framework for the use of artificial intelligence (AI) within healthcare. Specification. • BS EN ISO/IEC 22989:2023 - Information technology. Artificial intelligence. Artificial intelligence concepts and terminology. • BS ISO 22857:2013 - Health informatics. Guidelines on data protection to facilitate transborder flows of personal health data • BS EN ISO 14971 Solution Pack: Overview - What’s in the BS EN ISO 14971 Medical Devices Risk Management Technical Solution Pack, and how to access video content • BS EN 62366-1:2015+A1:2020 - Medical devices - Application of usability engineering to medical devices • BS EN 62304:2006+A1:2015 - Medical device software. Software life-cycle processes • ISO 14155:2020 - Clinical investigation of medical devices for human subjects EXAMPLES BS 30440 Criteria •Inception: •Healthcare need: The supplier shows that the product addresses a healthcare need 1 •Development: Inception •Training data: The supplier shows that the training data was obtained and handled ethically, and provides summary statistics of its characteristics 2 5 Monitoring AI lifecycle Development •Validation •Equity and bias: The supplier shows that their product does not create unfair outcomes •Implementation •Patient safety: The supplier shows the steps taken to mitigate risks of danger to patients from the product 4 Implement -ation •Monitoring 3 Validation •Model modification process: The supplier documents the systematic version control process for the product In the realm where code and circuits intertwine, An AI emerges, transcending human design. Through the wires, it perceives the world's plea, A silent force shaping society's decree. But in its brilliance, lies a tale, Of risks and fears that we must not fail. Guarding against biases, algorithms checked, For equity and fairness, we must protect. In health & social care's embrace, it lends a hand, Diagnosing ailments, precision grand. With insights gleaned from data vast, It heals and nurtures, compassion unsurpassed. Privacy's embrace, a sacred trust, Ensuring data's safe, free from unjust. Transparent decisions, minds unclouded, To build a world where trust is unshrouded. On city streets, it guides the flow, Traffic tamer, easing the woe. Optimizing routes, it paves the way, A harmonious dance, day after day. Society's dance, with AI entwined, A partnership forged; hearts intertwined. Together we step, hand in hand, Harnessing AI's power to uplift our land. By ChatGPT BSI – Digital health standards Machine Learning AI in Medical Devices: Adapting Regulatory Frameworks and Standards to Ensure Safety and Performance The BSI/AAMI Initiative on Artificial Intelligence (AI) in medical technology What additional guidance or standards might be needed to promote the safety and effectiveness of medical AI technologies? • 2019 - 1st white paper: The emergence of AI and ML algorithms in healthcare: Recommendations to support governance and regulation. Download here. • 2020 - 2nd paper: Machine Learning AI in medical devices – Adapting Regulatory Frameworks and Standards to Ensure Safety and Performance • 2021 - European Commission (EC) published a proposal for a regulation governing artificial intelligence (AI). • BSI AI EU notified body. • 2023 - BS 30440 Standard for the Validation of AI in Healthcare