Association between long-term hemoglobin variability and mortality in Korean adults: a nationwide population-based cohort study

Interpretable machine learning models for predicting 90-day death in patients in the intensive care unit with epilepsy

PURPOSE

This study aims to develop a machine learning-based model for predicting mortality risk in patients with epilepsy admitted to the intensive care unit (ICU), providing clinicians with an accurate prognostic tool to guide individualized treatment.

METHODS

We collected clinical data from clinical databases (MIMIC IV and eICU-CRD) of epilepsy patients 24 h after ICU admission. The clinical characteristics of ICU patients with epilepsy were carefully feature selected and processed. MIMIC IV as the training set and eICU-CRD database as the test set. Six models were developed and validated, and the best LightGBM model was selected by performance comparison and analysed for interpretability.

RESULTS

The final cohort comprised 429 patients for training and 1217 for testing. The training set exhibited a 90-day mortality rate of 9.32 %, and the test set had an in-hospital 90-day mortality rate of 4.10 %. Utilizing the LightGBM model, we achieved an AUC of 0.956 in the training set. External validation demonstrated promising results with accuracy of 0.898, precision of 0.975, AUC of 0.781, F1 score of 0.945, highlighting the model's potential for guiding clinical decision-making. Significant factors influencing model performance included the severity of illness, as measured by the OASIS score, and clinical parameters like heart rate and body temperature.

CONCLUSION

This study introduces a machine learning-based approach to predict mortality risk in ICU epilepsy patients, offering a valuable tool for clinicians to identify high-risk individuals and devise personalized treatment strategies, thus improving patient prognosis and treatment outcomes.

Iron- and protein rich diets may boost hemoglobin levels among informal electronic waste recyclers exposed to metals at Agbogbloshie, Ghana

While human exposure to metals may play a role in the pathogenesis of anemia, consumption of balanced diets may boost blood hemoglobin (Hb) levels in humans. Although informal electronic waste (e-waste) recycling processes have recently drawn attention as an important source of pollution, there is almost no empirical evidence on the relationship between diet, metals exposure and anemia among e-waste recyclers. Therefore, we evaluated possible ameliorating effects of diet on metal exposure related anemia, as measured by Hb levels of e-waste recyclers and a reference population in Ghana. This repeated measure study used data collected from e-waste recyclers (n=142) and a reference population (n=65) between March 2017 and October 2018. Stored whole blood samples were analyzed for the following metals; Cd, Pb, Rb, Tb, Tl, and Eu. Next, Hb levels were analysed using the URIT-810® semiautomatic biochemistry analyzer. Furthermore, a 48-hour dietary recall questionnaire was administered to assess dietary intake parameters such as protein, folate, carbohydrates, Fe, Ca, Mg, Se, Zn, and Cu. Ordinary regression models were used to estimate joint effects of metals and nutrients on Hb levels. At baseline, the mean Hb was lower among recyclers (12.99 ± 3.17 g/dL) than the reference group (13.02 ±2.37 g/dL). Blood Pb, Cd, Rb, Eu and Tb were associated with significant decreases in Hb levels of e-waste recyclers. Dietary intake of proteins and Fe was associated with concomitant increase in Hb levels of both groups as well as when analysis was restricted to recyclers. Despite the high exposure of e-waste recyclers to a myriad of metals, consumption of Fe-rich diets appears to ameliorate anemia and improved Hb levels (β=0.229; 95% CI: 0.013, 0.445; p=0.04). Therefore, the consumption of Fe and protein-rich foods may boost blood Hb levels in e-waste recyclers, even though exposure to high levels of metals is a predictor of anemia among this worker-group.

Association of hemoglobin variability with the risk of cardiovascular disease: a nationally representative retrospective cohort study from South Korea

Hemoglobin variability is known to increase cardiovascular mortality in chronic kidney disease, but the association of hemoglobin variability with the risk of cardiovascular disease (CVD) in the general population is yet unclear. This retrospective cohort study based on 'the South Korean National Health Insurance Service database' consisted of 198,347 adults who went through all three health examinations. Hemoglobin variability is defined as the average successive variability of three separate hemoglobin values from each health screening period. Participants were followed up for 6 years to determine the risk of coronary heart disease and stroke. We used multivariate Cox proportional hazards regression to examine the adjusted hazard ratios for CVD according to hemoglobin variability. Per 1 unit increase of hemoglobin variability, the risk for CVD (aHR 1.06, 95% CI 1.02-1.09) and stroke (aHR 1.08, 95% CI 1.03-1.13) increased significantly. The risk-increasing trend was preserved in the low-to-moderate risk group of CVDs (aHR 1.07, 95% CI 1.02-1.11). This result suggests that subjects with high hemoglobin variability who would otherwise be categorized as having low-to-moderate CVD risk may have higher risk of CVD than those with low hemoglobin variability.

Association between hemoglobin variability and incidence of hypertension over 40 years: a Korean national cohort study

Hemoglobin level determines blood viscosity and as hemoglobin level rises, blood pressure rises. However, hemoglobin level in individuals is not fixed and change in hemoglobin is affected by various clinical conditions. The purpose of this study is to investigate whether the hemoglobin variability affects the development of hypertension using Korean cohort database. This study was conducted with 94,798 adults (age ≥ 40 years) who visited the health screening in 2006 or 2007 (index year) and had at least 3 health screenings from 2002 to 2007. Hemoglobin variability was assessed by 3 indices of coefficient of variation (CV), standard deviation, and variability independent of the mean. Cox proportional hazard regression analysis was performed for each index of quartile groups (Q1–Q4). A total of 29,145 participants (30.7%) had the incidence of hypertension during a median follow-up of 7.4 ± 2.5 years. In the multivariable adjusted model, the hazard ratio and 95% confidence interval for incidence of hypertension of Q2, Q3, and Q4 compared with Q1 of hemoglobin variability CV were 1.014 [0.981–1.047], 1.064 [1.030–1.099] and 1.094 [1.059–1.131] respectively. The results were consistent in various sensitivity and subgroup analyses. This study showed that hemoglobin variability could be associated with hypertension development.