Learn to Discover (L2D): A Training Platform in Data Sciences and Machine Learning for Biomedicine and Health Researchers.

Digital skills need to improve to optimise competitive potential from accelerating applications of digital technology. Biological and medical science is becoming more automated to tackle bigger and more complex problems in discovery and applied health and medical sciences. The need for routine adoption of radically new ways of working and optimisation of their impact in this critical sector will only increase. We need action to re-skill the UK scientific workforce continuously since the deployment of digital skills offers very strong growth opportunities. The UK Government has committed to a technological future and to sectors that can deliver economic growth and has learnt lessons from elsewhere: 50% of all growth in the US economy over the last 50 years has come from the 5% of the workforce in STEM disciplines. Fusing the adoption of digital tech skills to the especially high-value STEM disciplines of discovery bioscience, biomedical and health sciences, where the UK already punches well above its weight, has tremendous potential for meaningful and measurable economic and social impact. However, tackling a digital skills crisis is not a trivial undertaking "Britain's chronic supply issues requires radical action. Working as a data expert requires knowing your maths, coding and computer science as well as problem solving, resilience and communication." (The Royal Society: - May 2019).

Since 2011 SysMIC (http://sysmic.ac.uk) - funded initially by the BBSRC - has taken advantage of the digital technology, internet access, and distant communication infrastructure widely available for most professionals to address half of this skills problem amongst active bioscience and health researchers. We delivered high-quality, e-learning and training in mathematical, computational and statistical methods. The Learn 2 Discover (L2D) project will combine the expertise of leading health and bioscience-facing computational and data scientists with the remote learning acumen and resources of SysMIC to solve the remaining part of the challenge. L2D we will deliver data science, machine learning and AI training in a highly accessible, flexible, modular format, suitable for a very wide range of starting expertise - including beginners - and study regimes. Our modules will draw on established real-world examples yet deliver widely applicable skills and general computational self-confidence for effective application well beyond the course. Delivery through the web offers resilience to disruptions of HE systems and leverages remote work and study competence developed across the R&D workforce in 2020 Participation of UK bioindustry stakeholders in the design of the programme will promote movement and sharing of talent between academic and commercial sectors. Collaboration and alignment of our modules with the work of UK centres of research excellence, infrastructure and resource networks will promote visibility, confidence and demand.

L2D will squarely address the digital productivity puzzle and promote knowledge exchange and its translation into impact in society and the economy and will offer opportunities for cross-sector spill over of benefits from training. Failure to respond effectively to the digital skills challenge is a major risk to business growth, innovation and broader societal development. A shortage in suitable digital skills persists in the UK labour market - research biomedical and health sciences is not an isolated case. Demand for workers with specialised data sciences and computational skills has been growing 6.5-fold faster than all other requirements. The best option is to nurture talent in the bio-, biomedical and health research sectors in situ with first class CPD of the sort proposed in L2D. A persistent digital skills wage differential makes impactful, broad digitals skills training attractive and very good value for money since over 75% of job openings at any skill level request digital skills.

In bio and health sciences data complexity and volume is now outstripping the compute-intensive physical sciences. Challenges of massive datasets, needs for real-time predictive modelling and computer execution of human intelligence-level tasks are growing rapidly. To address this, research and education leaders in the applications of data science and quantitative approaches to bio-, biomedical and health research will build on an outstanding track record to create a data science, ML and AI e-learning and training platform called Learn 2 Discover (L2D) to meet the needs of their wider peer community. L2D will develop specific, highly relevant skills for participant's immediate needs. Insights into the data challenge of other fields will narrow persistent knowledge gaps - for example, through L2D, drug developers might encounter more data aware clinicians and conversely laboratory scientists might appreciate the priorities of health practitioners and what data is gatherable.

L2D will deliver:
a) Core skills in data handling, supervised and unsupervised ML, working with complex, high dimensional data. ("live" within 4 months of start-up)
b) A portfolio of "use case" scenarios e.g. image analysis, digital pathology, neuroscience, drug development, molecular structure-function, genotype-phenotype relationships, synbio, cancer diagnostics and functional genomics using validated real-world data
c) Insight into advances in AI and Deep Learning e.g. various neural networks, reinforcement learning, autoencoders and natural language processing and more
d) Good practice in computing and data handling to address the reproducibility crisis
e) Web publishing for sharing predictors and code
f) Accessing cloud computing to spin-up flexible resource for large compute problems
g) Collaborative coding and project workshops with skilled tutors
h) Support materials and coaching for "local" trainers in universities and businesses to facilitate the uptake and adoption.