Machine learning versus logistic regression for the prediction of complications after pancreatoduodenectomy

Construction of risk prediction model for hypothermia during pancreaticoduodenectomy

Purpose

To investigate the factors influencing hypothermia during pancreaticoduodenectomy and establish and verify a prediction model.

Method

The clinical data of patients undergoing pancreaticoduodenectomy in Hunan People's Hospital between January 1, 2022 and October 15, 2022 were analysed. The patients were divided into a hypothermia group (n = 302) and a non-hypothermia group (n = 164) according to whether hypothermia occurred during surgery. A binary logistic regression model was used to analyse the independent risk factors for hypothermia in patients undergoing pancreaticoduodenectomy. A risk prediction model was established, and R software was used to plot a column graph. The predictive value of the model was evaluated using the receiver operating characteristic (ROC) curve.

Results

Among the 466 patients undergoing pancreaticoduodenectomy, 302 (64.81 %) had hypothermia, including 154 men and 148 women, with a median age of 58.6 (38-86) years. The binary logistic regression analysis showed that low body mass index (BMI), room temperature at the time of entry, intraoperative flushing fluid volume and peritoneal flushing fluid temperature were independent risk factors for intraoperative hypothermia in patients undergoing pancreaticoduodenal surgery (P < 0.05). A multivariate logistic regression analysis (backward logistic regression) was used to establish the prediction model. The area under the ROC curve was 0.927, P ≤ 0.001, the sensitivity was 0.921 and the specificity was 0.848, indicating good differentiation by the prediction model.

Conclusion

The nomogram constructed using four independent risk factors: BMI, room temperature at the time of entry, intraoperative peritoneal flushing fluid volume and intraoperative peritoneal flushing fluid temperature, has good predictive efficacy and good clinical application value for predicting intraoperative hypothermia in patients undergoing pancreaticoduodenectomy.

Optimizing prediction models for pancreatic fistula after pancreatectomy: Current status and future perspectives

Postoperative pancreatic fistula (POPF) is a frequent complication after pancreatectomy, leading to increased morbidity and mortality. Optimizing prediction models for POPF has emerged as a critical focus in surgical research. Although over sixty models following pancreaticoduodenectomy, predominantly reliant on a variety of clinical, surgical, and radiological parameters, have been documented, their predictive accuracy remains suboptimal in external validation and across diverse populations. As models after distal pancreatectomy continue to be progressively reported, their external validation is eagerly anticipated. Conversely, POPF prediction after central pancreatectomy is in its nascent stage, warranting urgent need for further development and validation. The potential of machine learning and big data analytics offers promising prospects for enhancing the accuracy of prediction models by incorporating an extensive array of variables and optimizing algorithm performance. Moreover, there is potential for the development of personalized prediction models based on patient- or pancreas-specific factors and postoperative serum or drain fluid biomarkers to improve accuracy in identifying individuals at risk of POPF. In the future, prospective multicenter studies and the integration of novel imaging technologies, such as artificial intelligence-based radiomics, may further refine predictive models. Addressing these issues is anticipated to revolutionize risk stratification, clinical decision-making, and postoperative management in patients undergoing pancreatectomy.

Machine learning approach to predict postpancreatectomy hemorrhage following pancreaticoduodenectomy: a retrospective study

BACKGROUND

Postpancreatectomy hemorrhage (PPH) is a rare yet dreaded complication following pancreaticoduodenectomy (PD). This retrospective study aimed to explore a machine learning (ML) model for predicting PPH in PD patients.

METHODS

A total of 284 patients who underwent open PD at our institute were included in the analysis. To address the issue of imbalanced data, the adaptive synthetic sampling (ADASYN) technique was employed. The best-performing ML model was selected using the PyCaret library in Python and evaluated based on recall, precision, and F1 score metrics. In addition to assessing the model's performance on the test data, bootstrap validation (n = 1000) with the original dataset was conducted.

RESULTS

PPH occurred in 11 patients (3.9%), with a median onset time of 22 days postoperatively. These minority cases were oversampled to 85 using ADASYN. The extra trees classifier demonstrated superior performance with recall, precision, and F1 score of 0.967, 0.914, and 0.937, respectively. Both validation using the test data and bootstrap resampling consistently demonstrated recall, precision, and F1 score exceeding 0.9. The model identified the peak value of C-reactive protein during the first 7 postoperative days as the most significant feature, followed by the preoperative neutrophil-to-lymphocyte ratio.

CONCLUSIONS

This study highlights the potential of the ML approach to predict PPH occurrence following PD. Vigilance and early interventions guided by such model predictions could positively impact outcomes for high-risk patients.

Machine learning versus binomial logistic regression analysis for fall risk based on SPPB scores in older adult outpatients

Objective

To compare the performance of the diagnostic model for fall risk based on the short physical performance battery (SPPB) developed using commercial machine learning software (MLS) and binomial logistic regression analysis (BLRA).

Methods

We enrolled 797 out of 850 outpatients who visited the clinic between March 2016 and November 2021. Patients were categorized into the development (n = 642) and validation (n = 155) datasets. Age, sex, number of comorbidities, number of medications, body mass index (BMI), calf circumference (left-right average), handgrip strength (left-right average), total SPPB score, and history of falls were determined. We defined fall risk by an SPPB score of ≤6 in men and ≤9 in women. The main metrics used for evaluating the machine learning model and BLRA were the area under the curve (AUC), accuracy, precision, recall (sensitivity), specificity, and F-measure. The commercial MLS automatically calculates the parameter range of the highest contribution.

Results

The participants included 797 outpatients (mean age, 76.3 years; interquartile range, 73.0-81.0; 288 men). The metrics of the current diagnostic model in the commercial MLS were as follows: AUC = 0.78, accuracy = 0.74, precision = 0.46, recall (sensitivity) = 0.81, specificity = 0.71, F-measure = 0.59. The metrics of the current diagnostic model in the BLRA were as follows: AUC = 0.77, accuracy = 0.75, precision = 0.47, recall (sensitivity) = 0.67, specificity = 0.77, F-measure = 0.55. The risk factors for falls in older adult outpatients were handgrip strength, female sex, experience of falls, BMI, and calf circumference in the commercial MLS.

Conclusions

The diagnostic model for fall risk based on SPPB scores constructed using commercial MLS is noninferior to BLRA.