Discovering new drugs, diagnostics, and biomarkers is a now a data science, generating a huge quantity of complex biological, chemical, clinical and safety information. Data mining, capture and analysis is essential for the processes of drug discovery, repurposing and development, the running of clinical trials, progression of precision medicine and improvements to patient care through genetically informed drug prescribing.
In the UK, the NHS is uniquely capable of collecting and making available high-quality data at a population scale. The opportunities that arise from this vast quantity of data have been widely acknowledged and only two months ago the government launched a plan to harness artificial intelligence as a weapon in the fight against cancer.
Over the course of two days last week (July 2nd-3rd 2018), 111 hackers met at the site of the UK’s contribution to the Human Genome Project, to try to solve 5 challenges in the sphere of genomes and biodata in 2 days. They had to form multi-disciplinary teams and collaborate across coding, science and medicine to come up with prototype solutions to the thorniest issues in human genetics.
The group was made of postdocs, Ph.D. students, health care professionals, health tech innovators and budding entrepreneurs with backgrounds in genomics, biodata, user experience, design, drug development, engineering and ethics: all with the skills, experience and self-belief that they could make enough progress to convince the judges to pick their solution.
This brave band of hackers worked tirelessly to hack solutions, create demo prototypes and design business propositions against the challenges selected by the Medicines Discovery Catapult, AstraZeneca, Microsoft, Open Targets, and ARM-Atos-Cavium. They were supported by mentors from the Sanger campus, genetic businesses like Congenica and Illumina, and entrepreneurs. Winners for each of the challenges were decided by a panel of 17 judges, based solely on a 10-minute presentation and Q&A.
The Medicines Discovery Catapult challenge
The Medicines Discovery Catapult challenge was “How can we combine drug and genetic data to intelligently prescribe drugs?”, or in hacker speak, “prescriptions 2.0”. Drugs work differently for patients with different genetic make-up, leading to variable effectiveness and side effects. This causes large-scale inefficiencies which result in unnecessary costs to healthcare providers and critically adversely impacts a patient’s quality of life. A major reason for this is due to genetic variation which affects how individuals process a drug. However, current prescribing does not consider a patient’s genetic make-up.
The key aims for this challenge were to find a solution that enables clinicians to conduct more targeted genetic testing in order to assess a patient’s probable response to a given drug, according to their individual genes, prescribe more intelligently, and ultimately improve care and patient outcomes.
Challenge winners had to address the following data and genetic-heavy steps:
- Know which genes are currently associated with variable drug response
- Know which genes are associated with variable response of drugs close to approval
- Know the frequencies of the relevant gene variants in a population
- Know how many patients require the relevant drugs
- Estimate the number of patients likely to benefit from gene tests
- Define which tests are needed, for example for an individual case, a disease-focussed clinic or genetic testing panel, or a country
The Medicines Discovery Catapult challenge winners
The winning team was “Prescription 1.5”
Alex Brown (GSK), Michael Hughes (SciBite Ltd), Joseph Mullen (SciBite Ltd), Michaela Sptizer (Open Tragets EMBL-EBI) and Jan Wildenhain (AstraZeneca).
Prescription 1.5 secured top place by applying SciBite’s text mining technology to extract drugs, indications, gene/protein mentions and adverse events from prescription drug labels. This allowed the pharma-academic-start-up team to develop a proof of concept tool that would enable clinicians to access this information in a quick, easy and systemic way.
The team describes their solution as an “interruptive clinical decision support system (iCDSS)”, meaning that actionable information can be delivered directly to a clinician at the point of prescription. Furthermore, Prescription 1.5 is specifically designed to be easily integrated into any existing e-Prescribing software and IT systems.
Prescription 1.5 will next present their solution to the NHS Chair of Pharmacogenomics Sir Munir Pirmohamed for potential NHS implementation, and two UK start-ups are interested in the technology also.
Read an article from our winners for further detail on the winning solution >
Read an article from the Sanger Institute for further information on the event and winners >
