DVLA: Drivers Medical Casework System
A casework system which uses decision trees to assist DVLA staff in assessing the users' ability to drive based on responses to medical questionnaires inputted by the user.
Tier 1 Information
1 - Name
Drivers Medical Casework System
2 - Description
Drivers Medical Casework System
Purpose: To assist caseworkers on assessing drivers fitness to drive with notifiable medical conditions. This process implements a workflow for caseworkers and medical practitioners to follow How it鈥檚 Used: Question flows are created to capture medical information entered onto the system to align to predefined decision trees which will make recommendations to caseworkers who assess individuals fitness to drive. Medical professionals are employed to assess complex cases individually but all recommendations made by the system are reviewed and either approved or rejected by trained staff before decisions are made. Data Input: Personal details for case including medical details. Decision Support: All cases are reviewed by trained staff, the tooling is used to offer advice. Currently being used on very specific conditions only as a live pilot. Why it鈥檚 Used: Currently being evaluated to improve turnaround time for casework and improve road safety through quicker resolution of cases. Resource Allocation: Taken in as under BAU and project resource. Risk Reduction: If taken forward and developed for more complex medical casework has the potential to reduce human error and reduce turnaround times by giving advice to staff that will make the final decisions.
3 - Website URL
/guidance/general-information-assessing-fitness-to-drive
4 - Contact email
Tier 2 - Owner and Responsibility
1.1 - Organisation or department
Driver and Vehicle Licensing Agency
1.2 - Team
Drivers Medical Group
1.3 - Senior responsible owner
Service Manager 帽 Driver Services and Account
1.4 - External supplier involvement
Yes
1.4.1 - External supplier
Kerv Communications Compliance Practice Ltd (Kerv Digital) Made Tech NTT
1.4.2 - Companies House Number
Kerv Communications Compliance Practice Ltd (Kerv Digital) - 13489823 Made Tech - 06591591 (no Longer contracted) NTT - 03085018 (no longer contracted)
1.4.3 - External supplier role
Kerv Communications Compliance Practice Ltd, Made Tech, NTT provides / provided trouble shooting, development and support. Responses to medical questionnaires.
1.4.4 - Procurement procedure type
Call off through Crown Commercial service G Cloud Framework
1.4.5 - Data access terms
Contractual, access limited to support and trouble shooting
Tier 2 - Description and Rationale
2.1 - Detailed description
This tool is made up of decision trees that provide recommended decisions and next steps to DVLA staff by assessing the users ability to drive based on responses to medical questionnaires inputted by the user and medical professionsals. Typical decisions generated by decision trees are:
- To issue a full/short term driving licence
- The revocation of a licence
- Requirement to obtain more information from a medical professional
- Requirements to workflow to a caseworker /DVLA Doctor for consideration
2.2 - Scope
Whilst currently on a very limited number of cases for Diabetes medical conditions only, the intention eventually is to provide recommendations on a wide range of medical conditions that potentially impact an individual鈥檚 ability to drive and ultimately the DVLA decision.
The tool only provides the user with a recommended decision based on information added, medical advice and DVLA policy rules.
This tool is not used to assist staff on making any other DVLA decisions and does not make the final decision.
2.3 - Benefit
This tool reduces the time taken to conclude cases, improves the aim for cases to meet 100% accuracy, undertakes automatic updates to drivers system records, reduces the process time when compared to the existing casework system. The tool further reduces the risk of data breaches as it reduces the number of manual processes as well as reduces customer address discrepancies held between existing systems. This wil improve road safety concerns.
2.4 - Previous process
The existing casework systems is a recording system for cases with manual updates from documentation and no automated response or collation of data. It is not fully integrated with main corporate systems and is classed as legacy.
2.5 - Alternatives considered
Alternatives were considered but the product selected 鈥楧ynamics365鈥� fits into the organisations strategic approach. The tool offered further integration benefits with other Microsoft 365 tools deployment. The solution is based on updated decision tree rather than a machine learning tool.
Tier 2 - Decision making Process
3.1 - Process integration
Individuals must notify DVLA of a medical condition that could impact their fitness to drive. This tool is used once a driver (customer) has submitted a request to DVLA to either apply for a provisional licence, exchange a licence, notify us of a medical condition, renew their photocard licence after 10 years, replace a lost driving lience or update their drivers record . The driver would complete the application and submit their medical conditions. The tool would then kick in and this application would be applied to the pre-existing business rules based on medical advice from medical professionals and the Medical Advisory Panel which is populated into the decision tree. The tool would then generate a recommendation based on the inputs into the decision tree with the recommendations being evaluated by a caseworker or medical professional before any next steps or final licensing decision is made. This decision is then fed back to the driver as an outcome.
3.2 - Provided information
The tool presents on screen to the caseworker (staff member) recommendations based on medical information supplied by the customer and or medical professionals as to their ability to drive. Caseworkers then review this data and recommendation and make a decision based on the information.
If the caseworker does not agree with the decision this can be manually amended. At this point the caseworker will need to justify the reasons for the change. This will be a recorded as a drop-down option for the reason and mandatory free text box.
3.3 - Frequency and scale of usage
Currently limited to specific medical conditions to ensure accuracy of results. With a small number of staff testing the new tool Transactions for this phase limited to approximately 185 per week.
3.4 - Human decisions and review
The tool will present recommendations to the caseworker on:
- Next steps, for example; recommendation for which letter to send to the driver (customer) next
- Licensing decisions to the Drivers Medical caseworker.
The caseworker will be presented with these recommendations on next steps and licensing decisions for consideration. The human always makes the final decision which each action and will review the information when making this decision. The recommendation is helpful extra information to assist the caseworker in making a decision. The recommendation can be accepted or rejected. If rejected the case will be taken forward for further investigation by another member of staff mandated under the existing casework system.
3.5 - Required training
Staff are provided training on the tool, how it works and how to use the tool. Further support is also given by contractors all of which go through an induction process and are trained on the tool use like the permanent staff.
3.6 - Appeals and review
There are standard existing appeals procedures in place that will be used regardless of systems used.
The licence holder (customer) is sent a letter detailing the licensing decision, which includes details on their rights of an appeal.
If a licence holder has extra medical evidence thats not been considered as part of the licensing decision they can appeal in writing directly to DVLA for further consideration.
If a driver appeals with no extra medical evidence, they must contact their local magistrate court in England and Wales, or Sheriff鈥檚 court in Scotland, and notify DVLA of their intention to appeal.
This right in relation to automated decision making and profiling does not apply under the current iteration of Dynamics. Although Dynamics business rules (decision tree logic) will be applied to all notifications, the final licensing decision will be made by a human.
Tier 2 - Tool Specification
4.1.1 - System architecture
Dynamics 365 is the product used and is Azure cloud based. Due to the nature of the data held the system architecture is not published.
4.1.2 - Phase
Beta/Pilot
4.1.3 - Maintenance
Maintenance is via Microsoft standard regime e.g. Patching. The rulebase and decision tree are maintained and tested according to business rule changes. Further deployment will be based incremental development.
4.1.4 - Models
Rule based
Tier 2 - Model Specification
4.2.1 - Model name
Drivers Medical Casework (Dynamics)
4.2.2 - Model version
Live Cloud based Platform - D365 Customer Service
4.2.3 - Model task
The model takes the provided medical information and computes a recommended expected outcome and next steps to the caseworker
4.2.4 - Model input
Medical questionnaire completed by drivers (customers) with further feedback by medical professionals
4.2.5 - Model output
A recommended action to review and make a decision on
4.2.6 - Model architecture
Rule based decision tree, the rules are manually updated based on business rules set by DVLA policy and there is no automated learning.
4.2.7 - Model performance
Currently under evaluation as part of limited deployment
Rigorous testing of decision trees took place. This included ensuring the correct suggested decision is provided through all combinations, this compared against the answers submitted to the questions and the Assessing Fitness to Drive (AFTD) guidelines and legislation flow chart. This included amending the questions submitted and re-running the decision tree.
Manual accuracy checking on team - output from the Dynamics casework solution has continued to be checked by an SME team to ensure accuracy of decisions.
4.2.8 - Datasets
Medical conditions reported by individuals
4.2.9 - Dataset purposes
Synthetic information used to test application of business rules and prove decision trees. Results compared to manual process / evaluated by medical practitioners
Tier 2 - Data Specification
4.3.1 - Source data name
Drivers Data and reported medical conditions
4.3.2 - Data modality
Text
4.3.3 - Data description
Individuals personal details such as name and address. Driving licence details and entitlement. Reported medical conditions and medical advice from medical professionals.
4.3.4 - Data quantities
Approx 200 Gb is held as the database, casework data and system logs
4.3.5 - Sensitive attributes
Personal data, name, address and driving licence entitlements. Medical conditions supplied by drivers which is considered special category information.
4.3.6 - Data completeness and representativeness
This data set is accuract and representative as the information supplied is from the individual customer and they cannot miss information out.
4.3.7 - Source data URL
N/A Restricted tool with no public URL
4.3.8 - Data collection
Initial driving licence data collected under legal framework Road Traffic Act 1988, for the investigation into fitness to drive. This is source data and retained for entitlement to drive, the data is not further processed beyond the original purpose it was collected and not repurposed.
4.3.9 - Data cleaning
N/A
4.3.10 - Data sharing agreements
Data will be shared as detailed below within legal gateway or with the consent of the individual
4.3.11 - Data access and storage
DVLA Caseworkers and Medical Staff have access to this information as part of their role. Data Subject Access team to fulfil DSAR requests. Other parties include doctors, optometrists, occupational therapists and members of the Department For Transport Medical Advisory Panels. (Details are held within the Data Protection Impact Assessment (DPIA) for Drivers Medical (DM) CRM).
Staff have limited access to medical data and controls are in place to restrict access to medical data to only those that need it.
Tier 2 - Risks, Mitigations and Impact Assessments
5.1 - Impact assessment
Data Protection Impact Assessment (DPIA): An in depth DPIA has been completed and is regularly updated.
As the DPIA is a living document it has continued to be updated as and when new changes to Dynamics have been made. The most current iteration was updated in July 2024.
5.2 - Risks and mitigations
Risks Inaccuracy of recommendations Mitigated by experienced / qualified staff checking results, a parallel testing program is undertaken as decision trees are developed and tested. A full privacy risk assessment is contained within the DPIA.