My main interest lies in designing and building clinical products used by experts to augment or shape decision-making. I’m particularly interested in the zero to one phase of building digital diagnostics, the challenge of designing for interpretability, the process of bringing these new technologies through regulatory approval.
mindstrong: measurement-based mental health
As Clinical Product Manager for Mindstrong I have led the development of both patient and provider facing products, in support of Mindstrong’s new care delivery model. Mindstrong’s technology measures cognitive and emotional function through patterns in the way you type, tap and swipe your smartphone screen. These measures are used as an endpoint for clinical trials, for example, in the early detection of dementia, and in our clinical practice to help people with serious mental illness avoid hospitalizations and ER admissions. I work closely with design and data science to integrate Mindstrong’s digital biomarkers of brain function into both products, understand our clinicians’ workflow and build a new tele-health platform in support of the treatment innovations we are developing.
Karius: infectious disease diagnostics
I joined Karius, an infectious disease diagnostics start up, as a Product Manager in 2016 to learn traditional biotech diagnostics development. Karius uses next generation sequencing to detect over 2000 microbes from a single blood sample. I worked across laboratory, data science, design and regulatory to define both the product experience, sample processing operations, and support the analytical and clinical validation studies. We implemented a quality management system for CLIA and CAP audits, and I led the initial development of design controls for FDA and CE certification. Externally I joined the sales teams to meet with lab directors and infectious disease physicians, and led user research to support the definition of the company’s assay development roadmap.
radar-cns: predicting relapse in patients with depression, multiple sclerosis and epilepsy (2016)
RADAR-CNS is a European research consortium dedicated to developing passive, digital detection of relapse in depression, multiple sclerosis and epilepsy. The initiative is jointly led by King’s College London and Janssen Pharmaceutica NV, funded by the Innovative Medicines Initiative and includes 23 organisations from across Europe and the US. I joined RADAR as part of an academic secondment during my medical training, reporting into Prof Matthew Hotopf who leads the program. I started the effort to define the smartphone and wearable device sensors and technical specifications for the platform, working with a team of leading academics from across the consortium.
Outcomes Based healthcare: Passive patient reported outcomes (2015)
Outcomes Based Healthcare is the UK’s leading healthcare consultancy dedicated to helping healthcare providers and payers define and measure outcomes for value-based care. I led product development for Sense360, a novel digital tool for measuring health outcomes using passively gathered smartphone sensor data.
We gathered actively reported PRO data from a group of participants with diabetes, as well as passive, continuous sensor data from the phone. Using a machine learning approach we aimed to find clusters of sensor data that correlated with PRO scores, and develop a scalable, passive measurement of outcomes to support value-based commissioning of clinical services in the UK’s National Health Service. The product won a number of awards, including the Connected Mobility Solution of the Year 2016, European IT & Software Excellence Awards.
Skin AnalYtics: Machine Learning for Melanoma screening (2013)
While training to be a doctor I helped design and develop the first version of Skin Analytics’ smartphone application for the detection of melanoma. Leveraging behaviour change techniques from game design, Stanford’s BJ Fogg, and the UK’s Behavioural Insights team we developed a mobile experience to encourage people at high risk of skin cancer to self-examine. Skin Analytics machine learning algorithms could detect minor changes (one of the strongest predictors of melanoma) in a mole’s shape, colour or size over sequential smartphone images taken under varying light conditions. The product has since evolved with the development of the DERM (Deep Ensemble for the Recognition of Malignancy) and is currently being integrated into a number of clinical pathways with NHS Clinical Commissioning Groups (CCGs) across the UK.