Integrating Artificial Intelligence Into International Classification of Functioning, Disability, and Health Coding: Effectiveness of a Mobile Application for Patient Questionnaires

Not Applicable

Completed

• Conditions: ICF; Rehabilitation; Artifical Intelligence; Application

• Registration Number: NCT07021781
• Lead Sponsor: Tulip Medicine

Brief Summary

Mobile applications and artificial intelligence are increasingly integrated into medical practice, yet their impact on workflow optimization and diagnostic accuracy remains understudied. This study evaluates the effectiveness of the MedQuest mobile application in optimizing patient questionnaire processes and assesses the accuracy of AI-driven International Classification of Functioning, Disability and Health (ICF) coding in comparison to traditional clinician-based coding.

Detailed Description

The advancement of digital health technologies has significantly transformed clinical workflows, enabling the integration of mobile applications into routine medical practice. Smartphones and tablets equipped with specialized software have transformed methods of accessing medical information, communication between medical staff and patients, and approaches to healthcare delivery. As a result, there has been an improvement in clinical efficiency and optimization of doctors' working time.

Modern mobile applications in healthcare cover a wide range of functions, including medical reference books and drug databases, health monitoring applications, telemedicine services, and remote patient monitoring applications. Proper use of such applications demonstrates effectiveness in improving the quality of patient care and reducing appointment times. Moreover, the widespread adoption of smartphones among various age groups, including the elderly, contributes to the integration of mobile applications in the rehabilitation process. This opens up opportunities for effective control of treatment and rehabilitation processes in a remote format, which is particularly relevant for rural areas and regions with a shortage of medical personnel.

At the current stage, one of the key challenges faced by doctors is limited patient appointment time, especially in conditions of staff shortages and high load on the medical system. In Kazakhstan, general practitioners are allocated 15 minutes per patient appointment, while narrow specialists have 20 minutes. During this short period, the doctor must conduct patient interviews, perform examinations, and complete all necessary medical documentation. This significantly complicates the possibility of in-depth patient assessment and increases the risk of medical errors, especially when there is insufficient time for comprehensive clinical decisions.

Of particular interest in this context are developments using artificial intelligence (AI) technologies. They have the potential to accelerate diagnostics, support decision-making, and improve diagnostic accuracy, particularly in disease classification according to the ICF. However, the efficacy of AI-driven tools in functional health classification remains underexplored, particularly in real-world clinical settings. Research in this area could provide important data on how well artificial intelligence handles ICF diagnoses and how useful its recommendations can be for doctors.

The International Classification of Functioning, Disability and Health is a fundamental tool of modern medicine that has significantly expanded the traditional approach to patient assessment. Its fundamental value lies in its holistic view of human health, overcoming the limitations of the classical medical model, which focuses primarily on diagnoses and pathophysiological disorders. The ICF offers a universal standardised language for specialists worldwide, ensuring effective communication between representatives of different medical disciplines. In contrast to the disease-oriented diagnostic classification of ICD, the ICF focuses on the functional capacity of the individual, which is crucial for planning and evaluating the effectiveness of rehabilitation interventions. Its relevance also extends to various areas of clinical practice, including geriatrics, where it helps to comprehensively assess the condition of elderly patients; neurology, where it is used in stroke, brain injury and neurodegenerative diseases; orthopaedics and traumatology to assess functional limitations after injury; and psychiatry, paediatrics and social medicine (18-20). The ICF is particularly valuable in the context of health and insurance systems, where it serves as a basis for decisions about coverage and resource allocation.

The choice of MedQuest as a technological platform for our research is dictated by a complex of factors that determine the effectiveness of a modern approach to medical questionnaires. First of all, there was an acute internal need to develop a specialized application, as our rehabilitation practice involves the use and interpretation of a significant volume of scales and questionnaires to assess the functional status of patients. In this regard, automatic assignment of ICF diagnoses with a qualifier significantly reduces diagnosis time and increases the objectivity of patient functional status assessment. Moreover, the mobile questionnaire format not only minimizes errors associated with manual data transfer but also makes the process more accessible to patients. It should be noted that the system effectively collects and processes responses, providing convenient analytical tools, while cloud storage improves information accessibility for medical specialist. Additionally, the transition to electronic questionnaires also reduces environmental impact by decreasing paper and other resource consumption. Undoubtedly, special attention should be paid to data security - mobile applications are equipped with protection features such as encryption and multi-level authentication, ensuring medical information confidentiality in accordance with regulatory requirements of HIPAA (USA), GDPR (EU) and, importantly, order of Kazakhstan "On Approval of Rules for Collection, Processing, Storage, Protection and Provision of Personal Medical Data by Digital Healthcare Subjects". It is worth emphasizing that electronic questionnaire completion allows patients to think through their answers more carefully in a calm environment, without feeling time pressure, which, in turn, contributes to obtaining more accurate data and reducing stress associated with medical facility visits. Parallel to this, the capability for remote monitoring and patient interaction through mobile applications allows doctors to track condition dynamics without frequent personal visits, which is especially useful for patients with chronic conditions and those living in remote or hard-to-reach areas. At the same time, the application of artificial intelligence opens possibilities for identifying non-obvious correlations and formulating personalized recommendations, while ensuring a high level of data protection through encryption and authentication.

Given the relevance of the problem and insufficient study of the issue, this research aims to comprehensively evaluate the effectiveness of the MedQuest mobile application compared to traditional paper-based questionnaire methods, with two primary outcomes: the impact on questionnaire completion time during physician appointments and the accuracy of AI-generated ICF coding compared to coding performed by qualified medical specialists. Secondary outcomes include user satisfaction among clinicians and ICF coding consistency.

Study Timeline

• Study Start: 2024-11-25
• Primary Completion: 2025-02-26
• Study Completion: 2025-02-26
• Study First Submitted: 2025-05-29
• Status Verified: 2025-06
• Study First Submitted QC: 2025-06-11
• Last Update Submitted: 2025-06-11
• Study First Posted: 2025-06-15
• Last Update Posted: 2025-06-15

Recruitment & Eligibility

• Status: COMPLETED
• Sex: All
• Target Recruitment: 185

Inclusion Criteria

1. Aged 18 and older.
2. Owned a smartphone (iOS or Android operating system) with internet access.
3. Deemed by the investigator to be able to understand and comply with the study requirements.
4. Provided a signed and dated written informed consent form, along with any necessary personal data processing permissions, before any examination procedures.

Exclusion Criteria

1. Patients with severe cognitive impairments that would prevent them from understanding and completing the questionnaires independently
2. Patients with severe visual impairments that would interfere with their ability to use mobile applications
3. Patients who declined to participate after being informed about the study protocol

Study & Design

• Study Type: INTERVENTIONAL
• Study Design: PARALLEL

Primary Outcome Measures

Name	Time	Method
Accuracy of AI-generated ICF coding compared to clinician coding	At study completion, approximately 3 months	Agreement between International Classification of Functioning, Disability, and Health (ICF) codes generated by an AI system and codes from trained medical professionals. Agreement was assessed using the quadratic weighted kappa coefficient across 22 ICF domains covering body functions, body structures, activities and participation, and environmental factors.
Time required to complete and process questionnaire	Baseline	Comparative analysis of the total time spent on completing standardized questionnaires between the control group (paper-based) and the experimental group (mobile app). In the control group, this includes both the time it takes for patients to complete the questionnaire and the time it takes for the medical staff to digitize the data. In the experimental group, only the time it takes to complete the questionnaire is measured, as the digital format eliminates the need for additional data processing.

Secondary Outcome Measures

Name	Time	Method
Clinician Satisfaction with the MedQuest Mobile App	At study completion, approximately 3 months	The SUS assesses 10 parameters including system complexity, ease of use, need for technical support, integration of features, consistency, learnability, confidence in use, and required knowledge on a 5-point Likert scale.
AI Performance Metrics for ICF Code Classification	At study completion, approximately 3 months	Detailed evaluation of artificial intelligence system performance in ICF coding through sensitivity (recall) and specificity analysis for each individual ICF qualifier code (0-4). Performance metrics include per-code sensitivity and specificity values, micro-averaged metrics (weighted by instance frequency), and macro-averaged metrics (unweighted average across all codes). Sensitivity measures the AI's ability to correctly identify true positive cases, while specificity measures its ability to correctly identify true negative cases. Micro-averaging provides overall system performance weighted by code frequency, while macro-averaging gives equal weight to all qualifier levels regardless of prevalence.

Trial Locations

Locations (2)

"Green Clinic" LLC; 🇰🇿 Astana, Kazakhstan

"National Scientific Oncological Center" LLC; 🇰🇿 Astana, Kazakhstan