Daily ReportSep 19, 2026
Anesthesiology, September 19 edition
We read 11 papers and selected 3.
Summary
The most impactful studies evaluated a non-invasive perioperative thermal-monitoring platform, an explainable machine-learning model for one-year mortality after geriatric hip arthroplasty, and the analgesic effectiveness of pericapsular nerve group block. Together, these studies advance perioperative monitoring, risk stratification, and opioid-sparing regional anesthesia, although clinical adoption requires external validation or randomized confirmation.
Research Themes
- Non-invasive perioperative monitoring
- Explainable machine learning for perioperative risk prediction
- Opioid-sparing regional anesthesia
Selected Articles
1. Clinical applicable skin-interfaced thermal guiding sensor for warning perioperative abnormalities in core body temperature and blood perfusion rate.
This study developed a compact, skin-interfaced thermal platform capable of estimating core temperature, tissue perfusion, thermal conductivity, and surface heat flux without invasive probes. Phantom and human experiments demonstrated errors within ±10% or ±0.1 °C, and perioperative measurements agreed well with invasive esophageal temperature monitoring during laparoscopic surgery.
Impact: The platform addresses a clinically important monitoring gap during anesthetic induction, when invasive temperature monitoring may not yet be available. Its multimodal, non-invasive capability could improve early detection of perioperative thermal and perfusion abnormalities.
Clinical Implications: The device could supplement or, after appropriate validation, provide an alternative to invasive perioperative temperature monitoring, particularly during induction and in settings where invasive probes are impractical. It may also support individualized thermal management, but accuracy in diverse patient populations and during hemodynamic instability must be established.
Key Findings
- The sensor estimated tissue thermal conductivity, blood perfusion, core temperature, and surface heat flux using thermal-guiding materials and real-time heat-transfer modeling.
- Phantom and human-subject experiments showed measurement errors within ±10% or ±0.1 °C.
- During laparoscopic surgery, estimated core temperature agreed well with invasive esophageal measurements, and long-term monitoring showed correlation greater than 0.85 with a conventional dual-heat-flux sensor.
Methodological Strengths
- The technology was evaluated across phantom experiments, perioperative human measurements, and longer-term monitoring during sleep, exercise, and rest.
- Agreement was assessed against invasive esophageal temperature monitoring and a conventional dual-heat-flux sensor.
Limitations
- The available abstract does not establish performance across large, diverse perioperative populations or during severe hemodynamic instability.
- The study demonstrates measurement agreement but does not show that sensor-guided monitoring improves clinical outcomes.
Future Directions: Future studies should perform prospective multicenter validation against reference thermometry, evaluate accuracy during shock and vasopressor use, assess skin tolerance and device robustness, and test whether sensor-guided interventions reduce hypothermia, hyperthermia, or perfusion-related complications.
Continuous monitoring of core temperature and blood perfusion is essential during general anesthesia, yet current workflows leave a critical gap during induction, before invasive probes can be placed. We present a compact, skin-interfaced thermal sensing platform for non-invasive monitoring throughout the anesthetic process. The device combines thermal-guiding materials with real-time heat-transfer modeling and inversion algorithms to estimate tissue thermal conductivity, blood perfusion, core temperature, and surface heat flux.
2. Explainable artificial intelligence for predicting mortality in geriatric patients undergoing hip arthroplasty: Machine learning analysis using national health insurance data.
Using claims data from 17,290 patients aged 65 years or older, the study developed an explainable random-forest model for one-year mortality after hip arthroplasty. The model achieved an area under the curve of 74.6%; age, transfusion, male sex, dementia, socioeconomic status, cancer, heart failure, chronic kidney disease, and other comorbidities were prominent contributors.
Impact: The study combines a large national dataset with explainable artificial intelligence, making mortality risk factors more interpretable for clinicians. It provides a scalable framework for perioperative risk stratification, although the model remains internally validated and predictive associations should not be interpreted causally.
Clinical Implications: The model may help identify geriatric hip arthroplasty patients who require intensified preoperative optimization, transfusion planning, postoperative surveillance, and multidisciplinary support. It should not yet guide individual treatment decisions without temporal and external validation, calibration assessment, and evaluation of clinical utility.
Key Findings
- The analysis included 17,290 patients aged 65 years or older who underwent hip arthroplasty in 2019.
- The random-forest model achieved an area under the curve of 74.6% for one-year mortality prediction.
- The leading contributors included age, red blood cell transfusion, male sex, dementia, low socioeconomic status, solid tumors, congestive heart failure, chronic kidney disease, and general anesthesia.
Methodological Strengths
- The large national claims database provides substantial statistical power and broad population coverage.
- Random-forest variable importance and SHAP analysis provide both global predictor ranking and patient-level interpretability.
Limitations
- The model was developed from a single national claims dataset and lacks reported temporal or external validation.
- Claims data may incompletely capture clinical severity, functional status, laboratory values, frailty, and perioperative management details; the identified associations are predictive rather than causal.
Future Directions: Future work should conduct temporal and international external validation, compare calibration across risk groups, incorporate richer clinical and functional variables, and prospectively test whether model-assisted care improves survival, complications, resource use, or equity.
We performed machine learning analysis of population data to determine the major determinants of 1-year mortality among geriatric patients undergoing hip arthroplasty. The data of 17,290 patients aged ≥ 65 years who underwent hip arthroplasty in 2019 were extracted from the Korea National Health Insurance Service claims database. With 1-year mortality as the dependent variable, random forest variable importance and Shapley Additive Explanations (SHAP) were used to identify the major predictors among 31 included predictors and their associations with mortality.
3. Postoperative analgesic outcomes of fascia iliaca compartment block versus pericapsular nerve group block in hip fracture surgery: a retrospective observational study.
In 186 patients undergoing hemiarthroplasty for hip fracture, PENG block was associated with substantially lower pain scores at 6 and 24 hours than fascia iliaca compartment block. PENG block also reduced 24-hour tramadol use by approximately 104 mg and improved QoR-15 recovery scores, with associations persisting after adjustment for age, sex, and ASA physical status.
Impact: The study addresses an immediate perioperative problem in a vulnerable population and reports clinically meaningful reductions in pain and opioid exposure. Its findings support PENG block as a promising component of multimodal analgesia, while the retrospective design requires confirmation in randomized trials.
Clinical Implications: PENG block may be considered as an opioid-sparing regional analgesic option for hip fracture hemiarthroplasty, particularly when early mobilization and recovery quality are priorities. Local expertise, block safety, motor effects, and comparative effectiveness in diverse patients should be considered before routine adoption.
Key Findings
- The retrospective cohort included 186 adults: 90 received fascia iliaca compartment block and 96 received PENG block.
- PENG block was associated with lower VAS pain scores at 6 hours and 24 hours, with mean differences of −1.63 and −1.11, respectively, both with p<0.001.
- Twenty-four-hour tramadol consumption was 103.91 mg lower and QoR-15 scores were higher in the PENG group.
Methodological Strengths
- All eligible adult patients at the study center during the defined period were included, reducing selective enrollment within the cohort.
- Multivariable linear regression adjusted for age, sex, and ASA physical status, and outcomes included pain, opioid use, and patient-reported recovery.
Limitations
- The single-center retrospective design permits confounding by indication and other unmeasured differences between block groups.
- The sample size was modest, and the study did not provide randomized allocation, multicenter validation, or longer-term functional outcomes.
Future Directions: Prospective multicenter randomized trials should compare PENG block and fascia iliaca compartment block using standardized multimodal analgesia, assess motor blockade and block-related complications, and measure mobilization, delirium, length of stay, chronic pain, and opioid use beyond 24 hours.
BACKGROUND: Hip fracture surgery is associated with substantial postoperative pain, delayed mobilization, and increased opioid requirements, particularly in older adults. Ultrasound-guided regional nerve blocks are increasingly used as components of multimodal analgesia. This study compared postoperative analgesic outcomes between fascia iliaca compartment block (FICB) and pericapsular nerve group (PENG) block in patients undergoing hemiarthroplasty for hip fracture. METHODS: This retrospective single-center cohort study included all 186 eligible adult patients who underwent hemiarthroplasty for hip fracture under spinal anesthesia between October 2022 and October 2024.