Recruiting the brightest minds to serve the nation’s health is a critical first step. For Health Education England (HEE), however, this marks the beginning of a significant investment.
Over seven years, HEE dedicates considerable resources, sometimes as much as £250,000, to nurturing the careers of future doctors. Keeping these talented people in the NHS is vital, so reducing trainee turnover is a major strategic priority.
To achieve this, HEE recognised the need to move beyond broad workforce analytics and pinpoint the crucial moments in a doctor’s training where targeted support could make a tangible difference. This is where collaboration with digital transformation consultancy Transform began.
The teams undertook an investigation of HEE’s extensive data resources, with the objective to foresee potential issues, thereby nurturing a more supportive and sustainable training landscape.
This considerable undertaking led to a detailed examination of more than 10 million records sourced from various NHS and HEE databases, requiring a well-developed approach incorporating quantitative research, sound data management principles, and advanced predictive analytics.
The problem: The need for predictive attrition modelling
Once an employee has resigned, reversing that decision presents a significant challenge. This reactive approach often proves ineffective. The core problem was in identifying and addressing employee disengagement before they reach the point of resignation, a stage where interventions are far more likely to succeed in retaining valuable talent.
While high-level workforce data offered a broad perspective, it lacked the necessary precision to identify who might be about to resign and why to make an effective intervention.
To reduce turnover, HEE needed to anticipate when this might happen and implement timely interventions. This required a solution capable of sifting through a wealth of information to uncover subtle patterns and predict future trends at an individual level.
The solution: Building a predictive ecosystem
The joint solution focused on quantitative research, data management and predictive modelling, leading to:
- Building a unified data warehouse to process over 10 million records from multiple NHS and HEE systems, creating a scalable platform that data scientists across departments can use collaboratively.
- Reinforcing confidence in data by collating and cleaning data from different sources to develop a machine learning model that predicts attrition for every individual.
The results: 10 million records and 64% prediction accuracy
The developed machine learning model, which analysed 10 million data points linked to the junior doctor workforce, achieved over 60% accuracy in predicting attrition. As the project’s key outcome, the model assigned each junior doctor a score indicating their likelihood to leave, enabling targeted interventions.
Beyond these quantifiable results, the project also delivered significant impact in alignment with sustainable development goals, including:
- The model was built to ensure course satisfaction and overall happiness in health education.
- Designed to predict attrition without gender bias, the model ensured that trainees identified as likely to leave received equal support, irrespective of their gender.
- With the cost of training junior doctors and consultants ranging from £250,000 to £500,000, each instance where a prediction leads to successful intervention and prevents a trainee from leaving represents a substantial saving of public funds.
- The attrition model was specifically designed to support the progression of all trainees by providing both opportunity and educational resources.
By turning millions of data points into actionable insights, HEE is now better equipped to understand the needs of its trainee doctors, provide support, and foster a more sustainable career pathway.
This scalable platform and the knowledge gained through this partnership enables future data-driven innovations within HEE, ensuring the continued development and retention of vital healthcare professionals.
