Identification of self-care problem in children using machine learning.

Area of Science:

• Pediatric Health
• Artificial Intelligence
• Machine Learning in Healthcare

Background:

• Identifying self-care problems in children is complex and time-consuming for medical professionals.
• A global shortage of occupational therapists exacerbates the challenge of diagnosing these issues.
• Machine learning offers a potential solution to simplify and expedite the identification process.

Purpose of the Study:

• To employ machine learning models for identifying self-care problems in children.
• To evaluate the effectiveness of different classification algorithms on the SCADI dataset.
• To improve diagnostic accuracy and efficiency in pediatric self-care assessments.

Main Methods:

• Utilized the SCADI dataset for training and testing machine learning models.
• Addressed high dimensionality through data reduction techniques.
• Applied SMOTE (Synthetic Minority Over-sampling Technique) to balance the imbalanced dataset.
• Compared Naïve Bayes, J48, and Random Forest classification algorithms.

Main Results:

• The Random Forest classifier achieved the highest performance on the SMOTE-balanced data.
• Achieved a balanced accuracy of 99% in identifying self-care problems.
• The developed machine learning model surpassed the performance of existing expert systems.

Conclusions:

• Machine learning, particularly Random Forest with SMOTE balancing, is highly effective for identifying childhood self-care problems.
• This approach offers a more accurate and efficient alternative to traditional diagnostic methods.
• The study demonstrates the potential of AI to support healthcare professionals in pediatric assessments.