Citations

ABSTRACT Background Sunlight has profound impacts on physical and mental health, beyond vision, including effects on circadian rhythms, alertness, mood, and sleep. A family history of any mood disorders is strongly associated with psychiatric disorders including bipolar disorder. The purpose of this study was to evaluate the association between a family history of any mood disorder in patients with bipolar I disorder and solar insolation at varied international onset locations. Methods Data for this analysis were available from 5842 patients with a diagnosis of bipolar I disorder obtained at 83 collection sites in both hemispheres. This included 4752 patients from 71 collection sites in the northern hemisphere and 1090 patients from 12 collection sites in the southern hemisphere. Patient data variables were obtained from records or interviews. Solar insolation data were obtained from The National Aeronautics and Space Administration (NASA) Power database for each onset location, and the ratio of the mean monthly minimum/mean monthly maximum solar insolation was calculated. Typically, the ratio is largest near the equator (little yearly change in solar insolation) and smallest near the poles (large yearly change in solar insolation). Results A significant relationship was found between a family history of any mood disorder, the ratio of the mean monthly minimum/mean monthly maximum solar insolation, and gender. The odds of a family history of mood disorder increased as patient location nears the poles and decreased near the equator. Female gender also increased the odds of having a family history of a mood disorder. Conclusions This study highlighted the association between family history, solar insolation, and gender in international patients with bipolar I disorder. Given the profound effects of sunlight on human health, the family of patients with bipolar disorder who live in the same location with the same solar insolation, and especially females, may be at increased risk for a mood disorder.

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.