2024 IEEE International Conference on Digital Health (ICDH)
Optimizing Patient Care Pathways: Impact Analysis of an AI-Assisted Smart Referral System for Musculoskeletal Services 2024 IEEE International Conference on Digital Health (ICDH) | 979-8-3503-6857-4/24/$31.00 ©2024 IEEE | DOI: 10.1109/ICDH62654.2024.00021
Haider Raza1 , Dheeraj Rathee2 , Renato Amorim1 , Maria Fasli1 which can lead to a lack of coordination and communication between healthcare providers, and may result in gaps in care. c) Disparities in access [8]: Patients from certain demographics or geographic areas may face barriers in accessing specialist services, resulting in inequitable healthcare outcomes. d) Administrative burden [9]: Referral processes can be timeconsuming and complex, placing an administrative burden on healthcare providers and potentially delaying patient care. An AI-assisted referral system could help address some of these challenges by improving the accuracy and efficiency of referrals, ensuring that patients are referred to appropriate specialists and services, and reducing administrative burdens on healthcare providers. Around 100 million appointments in England alone were dedicated to an MSK complaint – all of which could be freed up if patients were given the choice of a physiotherapist as their first point of contact [10]. There is a specific operational and clinical need based on data that revealed it took over three months for a patient to access treatment, mainly because a GP referral form went straight to the central processing unit and contained minimal information [2]. Exploring issues inhouse, our clinicians reported a challenge that various formats of referral forms are being used to refer patients from a variety of sources, from different GP surgeries. This lack of consistency often creates backwards and forwards between the central processing unit and the person referring to get the information needed surrounding the injury, and then to determine the urgency of the triage. Due to this tedious process, each patient takes approximately 2-3 hours to triage. Because the waiting list was so long, some patients had often self-healed before the appointment but were still taking up a place in the system. Moreover, clinicians discovered that the referral forms frequently lacked comprehensive details regarding the onset of the problem or injury. As a result, they were often required to dedicate significant portions of the initial appointment to piecing together background information from the patient, even months after the incident occurred. The demand for a more streamlined system for physio services came from clinicians themselves, whose main drivers were a redefined referral pathway, an improved patient experience and increased staff efficiency. An AI-assisted referral system could help streamline this process and improve the efficiency and accuracy of referrals [11]. By analyzing patient data, such as medical records and test results, and applying machine learning algorithms, the system could identify patterns and
Abstract—The UK’s National Health Service (NHS) confronts critical challenges in patient referrals amidst rapidly growing musculoskeletal (MSK) care demands. Current systems contribute to extended waiting times, incomplete referrals, fragmented care, and access disparities. To address this, we proposed an AI-assisted Smart Referral System (SRS) that enhances accuracy, efficiency, and equity. The SRS integrates a patient web portal, AI triage for real-time recommendations, and a digital pathway for seamless data handling. The system aims to streamline, utilizing AI for data analysis and specialist recommendations, potentially reducing waits and administrative burdens. In this study, we examined data spanning from August 2022 to July 2023, covering a period of 12 months, to assess the influence of the SRS platform on service delivery, costeffectiveness, and time efficiency. The results showed a significant reduction in missing information, coupled with substantial time and cost savings both at the administrative and clinical levels. Index Terms—AI, Healthcare, Musculoskeletal, Recommendation Engine
I. INTRODUCTION The National Health Service (NHS) in the UK is facing several challenges when it comes to patient referrals [1]. One of the health conditions such as musculoskeletal (MSK) care demand is rising and significantly impacting individuals, employers, the NHS, and the economy [2]. MSK condition results in the loss of over 30 million working days annually in the UK, constituting up to 30% of GP consultations [3]. With an ageing population, the demand for MSK services is expected to rise, posing challenges, especially in socioeconomically disadvantaged areas and certain ethnic groups [4]. To manage huge demand, NHS referrals involve directing patients to specialists, which comes with several challenges. Some of the main problems include waiting times: patients may have to wait several weeks or months to see a specialist, which can lead to delays in diagnosis and treatment [5]. There are several reasons behind this unwanted delay: a) Inappropriate referrals [6]: sometimes, patients are referred to specialists who are not the best fit for their specific condition or needs, leading to further delays or ineffective treatment. b) Fragmented care [7]: Patients may be referred to multiple specialists or services, *This work was supported by Innovate UK Funding for Knowledge Partnership with Provide Community CIC. 1 H, Raza., R. Amorim, and M, Fasli. are with the School of Computer Science and Electronics Engineering, University of Essex, Colchester, CO4 3SQ, United Kingdom h.raza@essex.ac.uk 2 D. Rathee is with Provide Digital, 900 The Crescent, Colchester, CO4 9YQ, United Kingdom.
979-8-3503-6857-4/24/$31.00 ©2024 IEEE DOI 10.1109/ICDH62654.2024.00021
68
Authorized licensed use limited to: UNIVERSITY OF ESSEX. Downloaded on September 03,2024 at 06:26:44 UTC from IEEE Xplore. Restrictions apply.