The K-nearest neighbor algorithm predicted rehabilitation potential better than current Clinical Assessment Protocol

Area of Science:

• Health Informatics
• Machine Learning in Healthcare
• Clinical Decision Support Systems

Background:

• Clinical decision-making can be enhanced by advanced computational techniques.
• Machine-learning algorithms offer potential for data-driven clinical insights.
• Current clinical protocols may be surpassed by automated, data-driven methods.

Purpose of the Study:

• To explore the utility of an automatic, data-driven machine-learning algorithm for clinical decision making.
• To evaluate the performance of a machine-learning algorithm against existing clinical protocols.

Main Methods:

• Utilized a large database (N=24,724) of home care client health assessments (interRAI-HC) across eight Canadian regions.
• Compared the K-nearest neighbor (KNN) algorithm with the existing Assessment of Daily Living Clinical Assessment Protocol (ADLCAP).
• Defined rehabilitation potential as functional improvement or remaining at home after a 1-year follow-up.

Main Results:

• The KNN algorithm exhibited lower false positive rates in seven of eight regions and lower false negative rates in all regions compared to ADLCAP.
• Likelihood ratio statistics indicated KNN was uniformly more informative than ADLCAP.
• KNN demonstrated a more accurate prediction of rehabilitation potential.

Conclusions:

• Machine-learning algorithms, such as KNN, show significant potential to improve clinical decision-making processes.
• Data-driven approaches can enhance the accuracy and informativeness of clinical assessments.
• This study highlights the value of integrating advanced algorithms into healthcare.