Scoping review of clinical decision support systems for multiple sclerosis management: Leveraging information technology and massive health data

Abstract

Background and purpose: Multiple sclerosis (MS) is a complex autoimmune disease of the central nervous system, with numerous therapeutic options, but a lack of biomarkers to support a mechanistic approach to precision medicine. A computational approach to precision medicine could proceed from clinical decision support systems (CDSSs). They are digital tools aiming to empower physicians through the clinical applications of information technology and massive data. However, the process of their clinical development is still maturing; we aimed to review it in the field of MS.

Methods: For this scoping review, we screened systematically the PubMed database. We identified 24 articles reporting 14 CDSS projects and compared their technical and software development aspects.

Results: The projects position themselves in various contexts of usage with various algorithmic approaches: expert systems, CDSSs based on similar patients' data visualization, and model-based CDSSs implementing mathematical predictive models. So far, no project has completed its clinical development up to certification for clinical use with global release. Some CDSSs have been replaced at subsequent project iterations. The most advanced projects did not necessarily report every step of clinical development in a dedicated article (proof of concept, offline validation, refined prototype, live clinical evaluation, comparative prospective evaluation). They seek different software distribution options to integrate into health care: internal usage, "peer-to-peer," and marketing distribution.

Conclusions: This review illustrates the potential of clinical applications of information technology and massive data to support MS management and helps clarify the roadmap for future projects as a multidisciplinary and multistep process.

Keywords: artificial intelligence; big data; clinical decision support system; multiple sclerosis; precision medicine.

FIGURE 1

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta‐Analyses) flowchart. Articles were primarily identified with a search based on keywords and key authors (the last search was performed on 15 April 2024). CDSS, clinical decision support system; ECTRIMS, European Committee for Treatment and Research in Multiple Sclerosis.

FIGURE 2

Chronology of the clinical decision support system (CDSS) projects and their technological basis. The years indicate the first publication or the first release (if specified in the article). Expert systems (also known as knowledge‐based CDSSs) implement human knowledge into knowledge bases. CDSSs based on data visualization query and visualize the data of similar patients recorded in reference databases. Model‐based CDSSs implement mathematical predictive models. They are commonly termed “AI‐powered” or “AI‐driven.” *The “Prognosis for patients with RR‐MS” and “sNfL reference app” names were given by the authors of this review. AEDSS, Automatic Expanded Disability Status Scale; EBDiMS, Evidence‐Based Decision Support Tool in Multiple Sclerosis; GAMLSS, generalized additive model for location, scale and shape; GLM, generalized linear model; MS, multiple sclerosis; PHREND, Predictive Healthcare With Real‐World Evidence for Neurological Disorders; RR‐MS, relapsing–remitting MS; sNfL, serum neurofilament light chain; UISS‐MS, Universal Immune System Simulator (MS extension).

FIGURE 3

Contexts of usage of the clinical decision support system projects along the management of multiple sclerosis (MS). Peaks represent clinical relapses. Platform names are mentioned in parentheses. *The “Prognosis for patients with RR‐MS” and “sNfL reference app” names were given by the authors of this review. AEDSS, Automatic Expanded Disability Status Scale; EBDiMS, Evidence‐Based Decision Support Tool in Multiple Sclerosis; PHREND, Predictive Healthcare With Real‐World Evidence for Neurological Disorders; PRIMUS, Projections in Multiple Sclerosis; RR‐MS, relapsing–remitting MS; SMSreg, Swedish MS Registry; sNfL, serum neurofilament light chain; SP‐MS, secondary progressive multiple sclerosis; UISS‐MS: Universal Immune System Simulator (MS extension).

FIGURE 4

Advancement of the clinical development of the clinical decision support system (CDSS) projects. The roadmap has been inspired by the one conveyed by the DECIDE‐AI working group. Circles represent the reported development steps and are colored according to the accessibility of the CDSS. Multiple steps could be reported by a single article. Platform names are mentioned in parentheses. *The “Prognosis for patients with RR‐MS” and “sNfL reference app” names were given by the authors of this review. AEDSS, Automatic Expanded Disability Status Scale; EBDiMS, Evidence‐Based Decision Support Tool in Multiple Sclerosis; MD, medical device; MS, multiple sclerosis; MVP, minimum viable product; PHREND, Predictive Healthcare With Real‐World Evidence for Neurological Disorders; POC, proof of concept; PRIMUS, Projections in Multiple Sclerosis; RR‐MS, relapsing–remitting MS; SMSreg, Swedish MS Registry; sNfL, serum neurofilament light chain; UISS‐MS, Universal Immune System Simulator (MS extension).

FIGURE 5

Software integration approaches with a proposed classification of the covered clinical decision support system (CDSS) projects. Local applications are installed and contained in a single computation unit (desktop computer, smartphone, etc.). Applications may also be installed over a whole institution's network, typically with a client–server architecture. Web‐based applications do not require any installation by the customer. They are provided as “software as a service.” Standalone applications require manual data input, whereas interoperable applications have automatic data exchanges typically according to standard data formats. *The “Prognosis for patients with RR‐MS” and “sNfL reference app” names were given by the authors of this review. AEDSS, Automatic Expanded Disability Status Scale; EBDiMS, Evidence‐Based Decision Support Tool in Multiple Sclerosis; EHR, electronic health record; MS, multiple sclerosis; PHREND, Predictive Healthcare With Real‐World Evidence for Neurological Disorders; RR‐MS, relapsing–remitting MS; sNfL, serum neurofilament light chain; UISS‐MS, Universal Immune System Simulator (MS extension).

FIGURE 6

Software distribution options with a proposed classification of the covered clinical decision support system (CDSS) projects. MS Prognosis Simulation and MS Vista have no reported distribution. “Internal usage” is the case when a health care institution implements an algorithm as a CDSS with no external distribution. It relies on internal or outsourced software development. “Peer‐to‐peer” is an external distribution in a noncertified framework. For instance, software may be shared through URL links to access a web application or a download but not be certified for medical usage. “Marketing” distribution requires the CDSS to be certified for medical usage as a medical device or another qualification depending on the local regulation and the intended use. Although MSProDiscuss was intended for marketing distribution, it is currently unavailable through this channel. *The “Prognosis for patients with RR‐MS” and “sNfL reference app” names were given by the authors of this review. AEDSS, Automatic Expanded Disability Status Scale; EBDiMS, Evidence‐Based Decision Support Tool in Multiple Sclerosis; MS, multiple sclerosis; PHREND, Predictive Healthcare With Real‐World Evidence for Neurological Disorders; RR‐MS, relapsing–remitting MS; SaMD: Software as aMedical Device; sNfL, serum neurofilament light chain; UISS‐MS, Universal Immune System Simulator (MS extension).