JMIR Publications

ABSTRACT

Background:

The rapid digitalization in healthcare has led to the adoption of digital technologies, but limited trust in internet-based health decisions and the need for technical personnel hinder the utilization of smartphones and machine learning (ML) applications. To address this, automated machine learning (AutoML) is a promising tool that can empower healthcare professionals to enhance the effectiveness of mobile health (mHealth) applications.

Objective:

We employed AutoML to analyze data from clinical studies involving patients with chronic hand and/or foot eczema or psoriasis vulgaris who used a smartphone monitoring application. The analysis focused on itching, pain and Dermatology Life Quality Index (DLQI) development as well as app usage.

Methods:

After extensive data set preparation, which consisted of combining three primary data sets by extracting common but also by computing new features, a new pseudomised secondary data set with a total of 368 patients was created. Next, multiple ML classification models were built during AutoML processing, with the most accurate models ultimately selected for further data set analysis.

Results:

Itching development over 6 months was accurately modelled using the Light Gradient Boosted Trees Classifier model (LogLoss: 0.9302 for validation, 1.0193 for cross-validation, and 0.9167 for holdout). Pain development over 6 months was assessed using the Random Forest Classifier model (LogLoss: 1.1799 for validation, 1.1561 for cross-validation, and 1.0976 for holdout). Next, the Random Forest Classifier model (LogLoss: 1.3670 for validation, 1.4354 for cross-validation and 1.3974 for holdout) was used again to estimate DLQI development over 6 months. Finally, app usage was analyzed using the eXtreme Gradient Boosted Trees Classifier model (LogLoss: 0.6817 for validation, 0.7121 for cross-validation, and 0.6604 for holdout). Influential feature correlations were identified, such as therapy changes during the trial, disease activity at enrolment, and DLQI scores at follow-up. Increased app usage was associated with higher disease activity, physical job types, and higher BMI. Conversely, lower pain intensity at follow-up and nicotine use were linked to lower likelihood of app usage.

Conclusions:

This study provides valuable insights into the relationship between data characteristics and targeted outcomes in patients with chronic eczema or psoriasis, highlighting the potential of smartphone and AutoML techniques to improve chronic disease management and patient care.