Identifying Weekly Trajectories of Pain Severity Using Daily Data From an mHealth Study: Cluster Analysis.

Area of Science:

• Pain research
• Data science
• Health informatics

Background:

• Chronic pain patients exhibit variable pain severity over time.
• Existing research often clusters sparse pain data, necessitating analysis of daily data for weekly trajectories.
• Understanding week-to-week pain variability requires quantifying movement between pain severity clusters.

Purpose of the Study:

• To identify common weekly pain severity patterns (clusters) in individuals with chronic pain.
• To establish a foundation for developing predictive models of pain variability.
• To analyze transitions between identified pain clusters.

Main Methods:

• Clustering of weekly pain trajectories (n=21,919) using a k-medoids algorithm from mobile health study data.
• Sensitivity analyses to validate the cluster solution and assess data structure assumptions.
• Transition analysis to examine movement between consecutive weekly pain clusters.

Main Results:

• Four distinct clusters of weekly pain severity were identified: no/low, mild, moderate, and severe pain.
• Sensitivity analyses confirmed the robustness of the four-cluster solution.
• Demographic and condition-specific differences were observed, with younger individuals and those with fibromyalgia or neuropathic pain spending more time in severe pain clusters. Males spent more time in the no/low pain cluster. Movement between clusters primarily occurred between adjacent categories.

Conclusions:

• The identified clusters offer a concise representation of weekly pain experiences in chronic pain populations.
• These clusters are valuable for future research into between- and within-cluster variability.
• The findings support the development of accurate, patient-centered pain-forecasting tools.