At present, medical treatment has entered an era of big data, and more and more medical big data sets have been used to generate algorithm models to guide the quality improvement of medical care. Tens of millions of operations are carried out in China every year, and the obtained perioperative big data are expected to significantly optimize perioperative management and reduce complications after governance mining and algorithm application. Perioperative big data are a massive data set generated during the perioperative diagnosis and treatment of patients. Combined with big data governance technology and machine learning algorithm, it can be applied to preoperative assessment, adverse event prediction, depth evaluation of anesthesia, automatic drug delivery system, decision support system and ultrasonic image processing, so as to significantly improve the perioperative safety of surgical patients and reduce medical expenses, and thus improve the health level of the people and reduce the burden of national health economy.
- Article type
- Year
- Co-author
By comparing the differences in cerebral oxygen parameters with 3 brands of near-infrared spectrometers, INVOS 5100C（Medtronic）, EGOS-600（Aegean）and BRS-1（Casibrain Technology）, to investigate their consistency, correlation and ability to detect key desaturation events in healthy volunteers in a hypoxic condition.
From December 6, 2021 to January 18, 2022, 20 healthy volunteers were recruited from our hospital. They were randomly divided into group E（a single sensor of INVOS 5100C at the left forehead and that of EGOS-600 at the right）and group B（INVOS 5100C and BRS-1, as the other group）, with 10 subjects in each group. After the sensors were applied to the forehead of the subject, hypoxia was induced and reversed by adjusting the ratio of inhaled nitrogen to oxygen. Regional cerebral oxygen saturation（rSO2）, heart rate and pulse oxygen saturation（SPO2）were continuously monitored and recorded in the whole process. Wilcoxon test was used to assess the baseline value and changes in rSO2 between different devices. Bland-Altman and Deming regression analyses were employed to analyze the consistency and correlation between the 2 devices. Receiver operating characteristic（ROC）curve was plotted to analyze the performance of the different devices in detection of hypoxia.
All 3 devices responded similarly to the changes in inhaled oxygen concentration. ① The magnitude and rate of changes in rSO2 for EGOS-600 and BRS-1 were less than those of INVOS 5100C. The ΔrSO2 for INVOS 5100C, EGOS-600 and BRS-1 was 16%~34.15%, 6%~20.81% and 5%~23.81%, with an absolute median rate of change of（2.81~4.40）%/min,（1.50~2.14）%/min and（1.77~2.41）%/min, respectively. ②Bland-Altman analysis showed that the mean difference of EGOS-600 and BRS-1 to INVOS 5100C were 5.7 and 2.5 respectively. Deming regression displayed that the Person coefficient was 0.722 4 and 0.845 9, respectively（P<0.001）, which were in good consistency and correlation with INVOS 5100C. ③In terms of absolute values of 10%, 20% and 50% relative reduction in INVOS 5100C as the key interventional rSO2 thresholds, EGOS-600 detection rates were 90%, 48% and 40% and BRS-1 detection rates were 80%, 48% and 50%. ④The ROC curve results indicated the AUC value was 0.893 and 0.715 for EGOS-600 and BRS-1, respectively, and the cut-off threshold for hypoxia prediction was 59.3% and 56.0%, respectively.
The response of EGOS-600 and BRS-1 to oxygenation changes is similar to that of INVOS 5100C, and the correlation and consistency are good, which can accurately predict brain oxygen changes.
To develop an early postoperative delirium（POD）risk prediction model in patients undergoing cardiac surgery based on Extreme Gradient Boosting（XGBoost）and compare its prediction performance with that of a traditional logistic regression（LR）model in order to provide reference for early identification and timely intervention of the condition.
A case-control trial was conducted on 684 patients who underwent elective cardiac surgery under general anesthesia due to heart disease in the Second Affiliated Hospital of Army Medical University from March to July 2022. According to the outcome of their 3-day follow-up after operation, the patients were divided into delirium group（n=38）and non-delirium group（n=646）. The patients were randomly divided into a training set（479 patients）and a test set（205 patients）at a ratio of 7∶3. LASSO regression analysis was used to screen out important variables related to POD. LR and XGBoost were employed to construct the prediction models. The area under receiver operating characteristic curve（ROC-AUC）of the prediction models and the sensitivity and specificity under the optimal threshold were calculated and the prediction performance of different models was compared.
The 3-day postoperative delirium rate of the patients undergoing cardiac surgery was 5.56%. Compared with the non-delirium group, the patients in the delirium group were older（P<0.05）, had a higher proportion of diabetes（P<0.05）, and lower preoperative systolic blood pressure（P<0.05）and postoperative sleep score（P<0.05）. But there were no statistical differences in other indicators（P>0.05）. Then finally, 5 variables, including age, preoperative peripheral oxygen saturation, preoperative regional cerebral oxygen saturation, preoperative systolic blood pressure and postoperative sleep score, were included for modelling. The AUCs of LR and XGBoost models were 0.732（95%CI: 0.43~1.000）and 0.659（95%CI: 0.559~0.759）, respectively. LR model had a higher AUC value and better predictive performance, but, its sensitivity was 50%, lower than that of XGBoost model（67%）, and its specificity was 100%, higher than the other model（98.5%）.
The predictive performance of the prediction model based on XGBoost, an integrated learning algorithm, is not superior to the traditional LR model for postoperative delirium after cardiac surgery. LR model can well predict the occurrence of delirium after cardiac surgery and provide reference for early intervention and treatment. But XGBoost is more sensitive to the diagnosis of postoperative delirium.
To investigate the effects of controlled hypotension induced by sodium nitroprusside on regional cerebral blood flow and oxygenation in pigs.
Six healthy female Landrace pigs (aged 4~6 months, weighing 22~25 kg) were included in this study. Sodium nitroprusside was used for controlled hypotension under sevoflurane anesthesia (20%, 30% and 50% lower than the baseline of mean arterial pressure or mean arterial pressure (MAP), corresponding to mild, moderate and severe hypotension, respectively). Oxygen to see (O2C), a monitoring instrument, was used to measure regional cerebral blood flow (rCBF) and regional cerebral oxygen saturation (rSO2). rCBF, rSO2, regional cerebral haemoglobin (rHb), heart rate (HR), peripheral oxygen saturation (SpO2) and incidence of cerebral hypoxia events were recorded under different MAP.
rCBF was significantly correlated with MAP (r=0.793) and also obviously correlated with rSO2 (r=0.843) in the whole hypotension process. rSO2 had a linear correlation with rHb, MAP and rCBF (R2=0.853, P<0.05), respectively. And the order of their linear correlation with rSO2 is rHb>MAP>rCBF. During mild hypotension, rCBF and rSO2 were increased by 7.4% (P<0.05) and 2.5% (P<0.05), respectively compared with the baseline, and the increase of rCBF was more remarkable. During moderate hypotension, compared with the baseline, rCBF was decreased by 0.9% (P>0.05), while rSO2 was decreased by 5.8% (P<0.05). During severe hypotension, rCBF and rSO2 were decreased by 7.6% (P<0.05) and 12.1% (P<0.05) respectively, and there was no cerebral hypoxia event during mild to severe hypotension. From 5 to 15 min after drug withdrawal, both rCBF and rSO2 were increased by 8.4% and 3.7%, respectively when compared with the baseline (both P<0.05).
Controlled hypotension with sodium nitroprusside is beneficial to maintain cerebral perfusion and cerebral oxygenation, but mild hypotension can cause an increase in cerebral blood flow, which is disadvantageous to patients with increased intracranial pressure.
京公网安备11010802044758号