Can a deep learning model based on intraoperative time-series monitoring data predict post-hysterectomy quality of recovery?

Table 1 Type and nature of the data used in modeling

Type of data	Nature of data
Preoperative data
Demographics (Age, height, body weight, and BMI)	Numerical data
ASA physical status classification	Categorical data
Anesthesia-relevant history (general anesthesia, spinal anesthesia, nerve block or local anesthesia, postoperative nausea and vomiting, and motor sickness)	Categorical data
Comorbidities (psychiatric disease, neurologic disease, hypertension, cardiovascular disease, pulmonary disease, endocrinologic disease, renal insufficiency, and digestive disease)	Categorical data
Laboratory results (hemoglobin, hematocrit, and creatinine)	Numerical data
Intraoperative intervention data^a
Anesthetic time	Numerical data
Propofol, remifentanil, and sufentanil^b	Numerical data
Crystalloid, urine output, and blood loss	Numerical data
Intraoperative monitoring data
Respiratory rate and end-tidal carbon dioxide	Time-series data, captured from anesthesia machine
Heart rate, systolic blood pressure, diastolic blood pressure, mean arterial pressure, pulse oxygen saturation and body temperature	Time-series data, captured from vital sign monitors
Muscular tissue oxygen saturation	Time-series data, captured from NIRS-based tissue oximeter

BMI body mass index, ASA American Society of Anesthesiologists, NIRS near-infrared spectroscopy, iMODIPONV trial the intervention guided by Muscular Oxygenation to Decrease the Incidence of PostOperative Nausea and Vomiting (iMODIPONV) trial
^aFor the intraoperative intervention data, the totals of different variables for the entire surgery were used in modeling
^bSufentanil was the standardized opioid for pain control in the iMODIPONV trial

ISSN: 2047-0525