TEMP_ICU
Description
Two ICUs temperature data. The set provides the temperature history from T-720 min to T+240 min every 30 min together with Heart rate, Systolic and Diastolic Blood Pressure (T-380 min to T=0), and the hour of the day. Timone_ train.csv is from Hôpital de la Timone ICU (used for training 1237 patients) Timone_validate.csv is is from Hôpital de la Timone ICU (used for hyperparameter optimization 825 patients) Nord.csv is from Hôpital Nord ICU (used for validation 3084 patients) each files : Temp(t-xmin) Temperature x min before Temp(t+ymin) Temperature y min before Temp(t) Temerature at instant t hour(t) hour in the day at the instant t ARTs(t-xmin) Systolic Blood Pressure x min before ARTd(t-xmin) Diastolic Blood Pressure x min before
Files
Steps to reproduce
To eliminate potential artifacts (disconnections, arterial catheter purge or positional catheter, etc.), Data were initially filtered by taking the median value of the different parameters over a 30-minute interval. This method helps in reducing noise and outliers in the data. Temperature values below 30°C were considered erroneous and were not included in the analysis. The recorded core temperature values had a precision to the tenth of a degree. To obtain continuous data, the curves were smoothed using a Savitzky-Golay filter, a digital filter that can increase the signal-to-noise ratio without distorting the signal. The values thus extrapolated to four decimal places were closer to reality. The data from all included patients were preprocessed and then concatenated to obtain the data sets for the AI algorithms