Change detection on longitudinal data in periodontal research

Area of Science:

• Periodontology
• Biostatistics
• Dental Research

Background:

• Longitudinal data on attachment level (AL) or alveolar bone level are vital for tracking periodontal disease progression.
• Assessing changes in AL across multiple sites presents a challenge in periodontal research.

Purpose of the Study:

• To identify the most efficient method for detecting changes in attachment level (AL) in a general periodontal research setting.
• To compare the effectiveness of various statistical methods for sequential AL change detection.

Main Methods:

• Examination of existing methods: tolerance, running median, cumulative sum (cusum), and regression.
• Inclusion of change-point detection methods from statistical literature.
• Evaluation of methods for sequential decision-making across multiple periodontal sites.

Main Results:

• Several methods demonstrated equal effectiveness in detecting changes in attachment level.
• The regression method was identified as the most convenient among the effective methods.
• Detailed formulae and tables for method application are provided.

Conclusions:

• The regression method provides a practical and effective solution for analyzing longitudinal attachment level data in periodontal research.
• This study offers guidance for researchers in selecting appropriate statistical tools for monitoring periodontal disease progression.