The problem
What sparse results miss
A value at midnight and another at eight in the morning fix only the two ends of the story. Between them the organ may have held steady, dipped and recovered, or been sliding the whole time, and a chart that joins the two points with a straight line quietly picks one of those stories on the clinician's behalf.
The method
How each result is weighted
We estimate each biomarker with a modified Kalman filter, which keeps a running belief about where the value stands and how fast it is moving. When a new result arrives, the filter moves toward it by an amount that depends on how reliable that measurement is, so a clean sample pulls the estimate most of the way and a doubtful one only nudges it.
The same reliability sets how far the confidence band closes around the result, which is why the band in the plot above tightens sharply at some dots and barely at others.
The payoff
From estimate to forecast
Because the estimate carries a velocity as well as a position, it can be carried forward in time, with its uncertainty widening the further ahead it looks. That forward line is what tells us how long an organ has before it reaches a threshold, which is the question the next page takes up.