All Party Parliamentary Group (APPG) – Artificial Intelligence Evidence Meeting – Corona virus and the role AI can play in fighting it.
8th April 2020
Ross Edwards – IORMA Technology Director
Lord Clement-Jones and Stephen Metcalfe MP chaired the meeting as usual along with Professor Birgitte Andersen from The Big Innovation Centre
Professor Tim Spector
The first Speaker, Professor Tim Spector who helped create the No 1 health App in the UK, COVID Symptom Tracker by Zoe Global Ltd. and the website https://joinzoe.com/, that along with helping track the symptoms also maps the spread of the virus. Within the first week of release the app had 1 million downloads and 2 million by the second, it became the key element being that it collects data in real time.
One of Chris’s roles involves bringing together AI and data skills across the country, he spoke about four different areas.
1. Data – to make sure different types of data are used
2. Uncertainty – bias in the raw data should be expected and compensated for
3. Appropriate methodology – to use a range of models and also not just AI
4. Collaboration – the benefits of events such as hackathons and 48 hr challenges and the emergence of Citizen Science
Professor Shannon Vallor, University of Edinburgh
Shannon’s role involves using AI to advance public health outcomes while engendering public trust.
She spoke of the trust that the public has or does not have in the way the government uses personal data and stated that transparency must be prioritised. Shannon pointed out that even GDPR might have gaps in it and mentioned that lifeboat ethics, where you prioritise the present, could set a dangerous president and that governments should appeal on moral grounds and as a trusted partner going forward.
Richard Dybowski, Oxford University, AI researcher
He is working on ‘policy optimisation in the context of a pandemic’.
Richard mentioned the use of AI to support clinical decisions and the use of knowledge maps which brings different scientific and medical papers together – the uncertainty of models and the benefits of using a range of models.
James Kingston, Deputy Director at Hatlab
James was concerned about not giving people control over their personal data.
Open research facilities the need for public trust and empowerment in the haste to get things done. He spoke of Health traffic lights that combine location data with peoples social data to give them risk scores based on who they come into contact with and that data once collected will not be going away after this crisis. He emphasised the importance of minimising the data needed and that deeply identifiable data would not be necessary.
Adam Riccoboni, Critical Future Ltd
Adam broke his section down to five points
1. Containment – Use of the traffic light system, MIT’s safepath app to protect personal data
2. Detection – use of AI in diagnostic tools
3. Drug discovery – speeding up the process of finding appropriate drugs
4. Patient care – the use of AI and at the triage stage and in the nurse robots used in Italy
5. Prediction – AI was used in the prediction of Covid 19 in Dec by analysing media stories around the world. He also mentioned that data needs to be available especially live health data from the NHS.
There seemed to be concern and division regarding the public’s rights to use and hold their own data and the need for large amounts of accurate real time data. Also, how can all the checks be put in place and at the same time not slow the process down.
There were concerns about some countries deliberately changing their data and this data is also being fed back into more models.
The emergence of Citizen Science where the public’s data and experience is being used, such as in the app by Zoe and Hackathons respectively. It was also noted some countries have been very aggressive in data collection and will be keeping that data for 6 years.