O'Grady, Gregory

ABSTRACT Background Body surface gastric mapping (BSGM) non‐invasively assesses gastric myoelectrical activity along with real‐time symptom reporting. In adults, the development of normative intervals has underpinned new explanatory phenotypes, aiding clinical decision‐making. This study established normative reference intervals for adolescent BSGM metrics. Methods Healthy adolescents aged 12–17 years with a BMI ≤ 35 kg/m 2 were recruited from New Zealand, Australia, and the United States. BSGM using Gastric Alimetry (Alimetry, New Zealand) involved a 30‐min fast, followed by a 480‐kcal meal, and a 4‐h postprandial recording. Reference intervals were calculated for four validated metrics: Principal Gastric Frequency (PGF), body mass index (BMI)‐adjusted amplitude, Gastric Alimetry Rhythm Index (GA‐RI, indicating rhythm stability), and the fed: fasted amplitude ratio (ff‐AR). Results were reported at the median and 5th and/or 95th percentiles as appropriate. Key Results A total of 107 participants (52.8% female, median age 14 [IQR 13–16], median BMI 20.1 [IQR 18.75–22.40]) with mixed ethnicities were included. No substantive correlations were observed between BSGM metrics and demographics or anthropometric data. Therefore, a single set of normative reference intervals was established. Median PGF was 3.06 cycles per minute; reference interval 2.72–3.37. Median BMI‐adjusted amplitude was 37.80 μV; reference interval 20.0–72.0. Median GA‐RI was 0.51; reference interval ≥ 0.22. Median ff‐AR was 2.12; reference interval ≥ 1.0. Conclusions and Inferences This study presents normative reference intervals for BSGM spectral metrics in adolescent populations, informing the interpretation of tests in research and clinical practice.

Background Predictive models employing machine learning algorithms are increasingly being used in clinical decision making, and improperly calibrated models can result in systematic harm. We sought to investigate the impact of class imbalance correction, a commonly applied preprocessing step in machine learning model development, on calibration and modelled clinical decision making in a large real-world context. Methods A histogram boosted gradient classifier was trained on a highly imbalanced national dataset of >1.8 million patients undergoing surgery, to predict the risk of 90-day mortality and complications after surgery. Class imbalance correction strategies including random oversampling, synthetic minority oversampling technique, random under-sampling, and cost-sensitive learning were compared to the natural distribution ('natural'). Models were tested and compared with classification metrics, calibration plots, decision curve analysis, and simulated clinical impact analysis. Results The natural model demonstrated high performance (AUROC 0.94, 95% CI 0.94--0.95 for mortality; 0.84, 95% CI 0.84--0.85 for complications) and calibration (log loss 0.05, 95% CI 0.04--0.05 for mortality; 0.23, 95% CI 0.23--0.24 for complications). Class imbalance mitigation (CSL, ROS, RUS, and SMOTE) did not improve AUROC or AUPRC but increased recall and F1 scores at the expense of precision and accuracy. However, these methods severely compromised model calibration, leading to significant over-prediction of risks (up to a 62.8 % increase) as further evidenced by increased log loss across all mitigation techniques. Decision curve analysis and clinical scenario testing confirmed that the natural model provided the highest net benefit. Conclusion Class imbalance correction methods result in significant miscalibration, leading to possible harm whenused for clinical decision making.