Evaluation of mathematic models to assess platelet kinetics

Area of Science:

• Hematology
• Biostatistics
• Medical Physics

Background:

• Accurate determination of mean platelet survival time is crucial for understanding platelet kinetics and diagnosing related disorders.
• Various mathematical models exist for calculating platelet survival time, but their comparative performance and reliability are not fully established.

Purpose of the Study:

• To compare the goodness of fit of twelve different mathematical models for calculating mean platelet survival time.
• To identify the most reliable models for estimating platelet survival curves in reference subjects and patients.

Main Methods:

• Platelets were labeled with [111In]oxine in 15 reference subjects and 54 patients.
• Twelve mathematical models, including linear (LN), exponential (EP, EX), weighted mean (WM, MWM), multiple hit (MH), Dornhorst (DH, DH10), Meuleman (ML, ML10), alpha order (AO), and polynomial (PO), were evaluated.
• Goodness of fit was assessed using % standard deviation (s.d.) and coefficient of variation.

Main Results:

• Models MWM, MH, DH, ML, AO, and PO demonstrated better goodness of fit (3.2-3.3% s.d.) compared to LN, EX, EP, WM, DH10, and ML10 (4.4-5.2% s.d.).
• The MH model's results differed significantly from DH, ML, and AO models, indicating that direct comparison of results from different models is not feasible.
• Model performance varied in sensitivity to noise, with coefficients of variation ranging from 7.9-12.0% for reference subjects.

Conclusions:

• The modified weighted mean (MWM) method and the multiple hit (MH) model are recommended for reliable estimation of mean platelet survival time.
• Utilizing at least two mathematical models provides increased confidence in the calculated mean platelet survival time.
• For very short platelet survival times, the alpha order (AO) model is specifically recommended.