Medical intelligence vs. artificial intelligence - How psychiatrists arrive at treatment algorithms

Area of Science:

• Psychiatry
• Computational Psychiatry
• Clinical Decision-Making

Background:

• Treatment guidelines offer algorithms but lack personalization for complex clinical practice.
• Estimating individual treatment response remains a challenge in psychopharmacology.
• The gap between guidelines and individual patient needs in psychiatric prescribing is not well understood.

Purpose of the Study:

• To explore how senior psychiatrists bridge the gap between treatment guidelines and individual patient needs.
• To identify heuristics used by experienced clinicians for personalized psychopharmacology.
• To understand the factors influencing treatment choices beyond established algorithms.

Main Methods:

• Qualitative analysis of interviews with 16 senior German psychiatrists.
• Analysis of heuristics used for personalizing psychopharmacological treatment recommendations.
• Examination of the hierarchy and factors influencing clinical decision-making.

Main Results:

• Psychiatrists employ a flexible hierarchy, moving from diagnosis to comorbidity, incorporating individual factors like age, gender, and preferences.
• Guideline adherence decreases with complexity, increasing reliance on clinical experience and intuition.
• Prescribers' choices are influenced by training, mentors, and clinical routine, with limitations in objective diagnostics and communication hindering algorithm improvement.

Conclusions:

• Experienced psychiatrists utilize personalized heuristics, often diverging from strict guidelines, to tailor psychopharmacology.
• Clinical experience and intuition play a significant role in treatment decisions, especially in complex cases.
• Critical adoption of AI in psychiatry is necessary, considering current limitations and the value of experienced clinical judgment.