Use of AI to improve care of people with mental health problems and empower them in their own care.

The goal of APPLEmh  (Adaptive Planning and raPid Learning in mental health) was to demonstrate a collaborative decision support and adaptive planning system for mental health services.  A key aim is to enable patient involvement in decisions that affect them.

  • The technical approach adopted uses the PROforma language to develop “executable care pathways” which can support the capturing and interpretation of relevant data, support decision-making and active plan management and carry out other tasks in an adaptive way. 
  • OpenClinical provides the platform for the development of simple prototype user interfaces. The content of the prototypes is drawn from a selected subset of the NICE guideline for the management of depression in adults. The prototypes accessible here are not appropriate for clinical use. 
  • An important part of the feasibility study was a series of patient consultation meetings which produced a set of requirements identified by participants. These were: (a) transparency of decision-making; (b) opportunities for patients to express preferences and concerns; (c) ability to challenge decisions and request changes; (d) reliable communication and information-sharing. A series of prototypes were developed iteratively with improvements made in response to the discussions at each meeting. 
  • Demonstrable outputs are a prototype clinical-view and a patient-view. The clinical view is based on the NICE guidelines, while the patient view is based on the requirements gathered from the patient consultation meetings.
  • In addition to these specific models, the project has also suggested potential for re-usable patterns in workflows. These include a generic pattern for active monitoring and collaborative decision-making, which we hope to validate in other mental health applications, and in patient-centred multidisciplinary care more widely.
  • From a technical point of view we believe that it will be possible to implement a more advanced prototype that is comprehensive enough to be piloted in practical settings.

APPLEmh phase 1 pathway models:

Clinical view



Other project Deliverables:

Pathway model explanation and documentation - PDF

Example patient narratives and data models - PDF

Use Cases - PDF

Roadmap - towards a Rapid Learning Health System - PDF


Phase 2

  • In the next phase of APPLEmh, the pathway models will be developed into a practical service implementation in which distributed processes work together to enable collaborative decision-making, leading to semi-automated plan updating and rapid learning.
  • A key conclusion is that the required service is not just an “app” for individual users, but a distributed set of interoperable services (agents) that are embedded in a larger multidisciplinary professional network supporting service users and carer
  • A rapid learning framework has not yet been demonstrated; this requires that the logic-based foundations of PROforma are extended to incorporate data analytics and machine learning techniques (see Fox, 2016 and

No documents found.

Submit an Application