eHealth tools

An overview of  eHealth assessment and analysis tools that I have designed and developed can be found here. Many tools were developed using Concerto, the Psychometrics Centre, University of Cambridge's open-source tool for developing electronic assessments and web-based statistical applications. 

These demos are hosted on servers which receive intermittent maintenance which can affect performance. Please report broken links here

PatientMosiac at MD Anderson

The goal of PatientMosiac is to collect diverse longtudinal multiomic data, including patient-reported outcomes (PROs), from 10,000 patients at MD Anderson. To help meet this goal we have developed a next-generation patient assessment tool that integrates adaptive testing algorithms, fuzzy search, and individualized feedback and guidance from ASCO.  We have seen a 50% improvement in patient retention throughout the study compared to systems without these sophistications and were able to capture 10% more information about patient symptoms, compared to static assessments. 

Immunotherapy Toxicitiy (IOTOX) Guidelines at MD Anderson

Immunotherapies have been transformational for treating  many cancers but are associated with potentially severe side-effects. We leveraged The Divison of Internal Medicine's rich experience in treating these rare toxicities  in the creation of a dynamic electronic IOTOX treatment guideline system. The system will soon be available for use around the world. 

ATLanTiC Quality of Life Assessment 

A computer adaptive test developed in collaboration with the WHO International Hub for Quality of Life research at the University of Manchester. It uses an algorithmic question selection procedure based on a user's previous responses to deliver assessments that are shorter and more reliable than paper-based questionnaires. You can read the accompanying journal article here

Quality of life prediction tool 

It is true that "All CATs are black in the dark" - once we deploy computer adaptive tests in to the 'real world' it can be difficult to judge how well they are really working as we lose the ground truth . This system allows users to rate model-based predictions about their quality of life as a means to validate the models. I was awarded the 2016 New Investigator Award at the 26th International ISOQOL symposium for my presentation of this new method. 

GMC Colleague Questionnaire open-text analysis  

Open-text information is routinely collected in many health services but it is often left unused. When qualitative data are collected in enormous quantities; human raters are an inefficient and expensive way to make sense of this information. We trained an ensemble of machine learning algorithms to classify open-text reports of doctors' performance with startling accuracy! 

Computer adaptive depression assessment using the CES-D 

Adaptive version of popular 'legacy' depression screening instrument in collaboration with Aiden Loe, doctoral candidate in psychology, University of Cambridge.

Machine Interpretation of Natural Text - Sentiment Analysis in the Concerto Environment (MINT-SAUCE) 

An example implementation of sentiment analysis using machine learning within Concerto. Used for teaching machine learning courses. 

Other tools developed elsewhere. 

Predictive World (Gibbons, Hoferkova, Barclay and Popov in collaboration with Sid Lee, Stink Digital and Ubisoft Entertainment) 

The Psychometrics Centre worked with industrial partners to develop a predictive online system to support the launch of Watch Dogs 2. Predictive World draws together 2.5 billion pieces of publicly-available data to make predictions about individuals; now and in the future. More than just a marketing tool, the system draws on epidemiological and psychological research evidence to allow users to explore the effects of different lifestyle behaviours including smoking and drinking on outcomes related to health, work, and leisure. 

Predictive world received over a million users in the first two weeks of release. The website was well-received by the internationally community and was awarded two prestigious design awards, the D&AD graphite pencil and the ADC Gold Award. 

Apply Magic Sauce (Kosinski, Stillwell and Rust) 

Apply Magic Sauce (AMS, to its friends) is a personalisation engine that accurately predicts psychological traits from digital footprints of human behaviour. AMS uses machine learning algorithms to predict intimate psychological traits from Facebook 'Likes' (papers here and here) and open-text. 

Aurei X Concerto Demo (Stillwell, Aurei) 

Demo of the Aurei platform for fully flexible emotional reporting.