ML-CDSS Implementation

A systematic review found that only 34.2% of CDSS tools have been used in clinical practice, reflecting a longstanding implementation gap. Implementing the ML-CDSS would require changes to established processes anchored in national guidelines and evidence-based best practice. Because the ML-CDSS relies solely on blood chemistry data, it cannot account for behavioural factors or comorbidities, which could produce recommendations that do not fully reflect a patient's true clinical state. Known barriers to CDSS adoption in healthcare include new administrative burdens, usability issues, and friction with existing clinical workflows. Whole-unit buy-in and formalised processes for recording ML-CDSS recommendations were identified as prerequisites for implementation, alongside extensive staff training. Clinicians noted that staff are deeply accustomed to existing protocols aligned with NICE guidelines, with nurses described as strictly adhering to written protocol and avoiding deviations. Reduced frequency of visits or blood tests enabled by the ML-CDSS could be perceived negatively by patients who value face-to-face interaction during treatment.