BLOGs

A PERSPECTIVE OF ARTIFICIAL INTELLIGENCE APPLIED TO MEDICAL SCIENCES BASED ON EVIDENCE

ARTIFICIAL INTELLIGENCE IN HEALTHCARE: FOCUS ON CLINICAL DECISION SUPPORT SYSTEM (CDSS) IN DIABETES

 

CURRENT SCENARIO

  • According to Ong Kl et al., diabetes forecast estimates that 1.31 billion people will have diabetes by 2030.  
  • This stage shall pose a significant burden on the healthcare system.
  • As per our current knowledge, Artificial Intelligence driven tools may offer a significant leverage point to diabetes sanitary services, at least in the following fields:
    • CDSS (especially in resource-limited settings)
    • Image analysis
    • Patient self-management
    • Patient engagement

 

CDSS IN DIABETES: THE EVIDENC

  • Although the evidence of Artificial Intelligence and data-driven systems in diabetes is still emerging, different studies have pointed out benefits in terms of improved health outcomes.  
  • Specifically, published data indicate that Artificial Intelligence tools improve glycemic control in people with diabetes, reducing fasting and postprandial glucose levels and A1C.

 

CDSS IN DIABETES: SPECIFIC LEVERAGE POINT

❖ The most practical usefulness of Artificial Intelligence in diabetes is as follows:

  • Complications prediction (hospitalization, macro and micro-vascular complications, among others)
  • Diagnosis of the disease
  • Risk prediction assistance

 

ISSUES TO FACE

❖ As identified by evidence, the most relevant topics to improve regarding Artificial Intelligence technologies are:

  • Wide diversity in the user´s interface
  • Many diabetes – CDSS do not reflect outcome improvements, particularly those related to longer-term clinical outcomes
  • Data accessibility
  • Regulatory compliance and ethics
  • Equity aspects
  • Bias prevention
  • Data governance
  • Poor data regions
  • Poor digital literacy (especially in older adults)
  • Limited internet access
  • Data security (cloud strategies to ensure data quality and transparency)
  • Trust generation

 

ADDRESSING THE FUTURE

  • In diabetes, a potential research field to afford is the machine-learning techniques to predict hypoglycemia.
  • In the short term, more sophisticated Artificial Intelligence based CDSS is required, not just rulebased approaches but complex learning systems supported.
  • Healthcare records digitization is an unmet need to facilitate data sharing among stakeholders. In other words, a robust digital infrastructure is a big necessity nowadays.
  • Strategies to facilitate older people's access to digital tools are highly required. 

 

FINAL REMARKS

  • We are assisting an extraordinary revolution in healthcare with rapidly evolving Artificial Intelligence technologies and challenging scenarios in clinician/scientist settings.
  • However, Artificial Intelligence should complement, but not replace, the human mind in critical areas such as clinical diagnosis, optimizing this process.
  • All stakeholders (physicians, patients, hospital systems, academia, etc.) must adapt to these new scenarios to capitalize on potential Artificial Intelligence benefits related to global health.