[Constructing an early-warning model for mortality risk in heat stroke patients based on Fisher discriminant analysis]

Objective: To establish an early warning model to assess the mortality risk of patients with heat stroke disease. Methods: The case data of patients diagnosed with heat stroke disease admitted to the comprehensive ICU of Shanshan County from January 2016 to December 2020 were selected. According to the short-term outcome (28 days) of patients, they were divided into death group (20 cases) and survival group (53 cases) . The relevant indicators with statistically significant differences between groups within 24 hours after admission were selected. By drawing the subject work curve (ROC) and calculating the area under the curve, the relevant indicators with the area under the curve greater than 0.7 were selected, Fisher discriminant analysis was used to establish an assessment model for the death risk of heat stroke disease. The data of heat stroke patients from January 1, 2021 to December 2022 in the comprehensive ICU of Shanshan County were collected for external verification. Results There were significant differences in age, cystatin C, procalcitonin, platelet count, CKMB, CK, CREA, PT, TT, APTT, heart rate, respiratory rate and GLS score among the groups. Cystatin C, CKMB, CREA, PT, TT, heart rate AUC area at admission was greater than 0.7. Fisher analysis method is used to build a functional model. Results: The diagnostic sensitivity, specificity and AUC area of the functional model were 95%, 83% and 0.937 respectively. The external validation results showed that the accuracy of predicting survival group was 85.71%, the accuracy of predicting death group was 88.89%. Conclusion: The early warning model of heat stroke death constructed by ROC curve analysis and Fisher discriminant analysis can provide objective reference for early intervention of heat stroke.

Relationship between periulcer mucosal inflammatory status and recurrent bleeding after successful treatment of gastric ulcer bleeding and establishment of an early warning model

BACKGROUND

Gastrointestinal bleeding tends to recur and may cause mortality, and there are multiple risk factors for gastroin-testinal bleeding. Therefore, comprehensive data is urgently needed in clinical practice to predict the adverse prognosis of peptic ulcer bleeding, in order to more accurately manage patients in a hierarchical manner.

AIM

To analyze the relationship between periulcer mucosal inflammatory status and recurrent bleeding after successful treatment of gastric ulcer bleeding, and establish an early warning model based on this.

METHODS

Three hundred and seven patients with gastric ulcer bleeding admitted to the Second People's Hospital of Lishui City from January 2020 to December 2022 were retrospectively selected. The situation and duration of gastric ulcer rebleeding after successful initial hemostasis were recorded. The differences in clinical indicators, inclu-ding periulcer mucosal inflammatory status, were compared between the recurrent bleeding group and the non-recurrent bleeding group, using a multifactor Cox regression model, to identify the factors that affect rebleeding and establish a nomogram model for predicting rebleeding.

RESULTS

Kaplan-Meier curve analysis showed that there was a statistically significant difference in the occurrence of gastric ulcer rebleeding during the follow-up period between groups with different levels of periulcer mucosal inflammation (P < 0.05). Periulcer mucosal inflammatory status, use of non-steroidal anti-inflammatory drugs, alcohol consumption, ulcer size, Helicobacter pylori (H. pylori) infection, and Forrest grade were identified to be factors affecting the recurrence of bleeding after successful treatment of gastric ulcer bleeding (P < 0.05). Multifactor Cox regression analysis showed that periulcer mucosal inflammatory status, use of non-steroidal anti-inflammatory drugs, alcohol consumption, H. pylori infection, and Forrest grade were independent risk factors for recurrent bleeding after successful treatment of gastric ulcer bleeding (P < 0.05). The discriminability and calibration of the nomo-gram developed for the prediction of rebleeding risk were good.

CONCLUSION

The more severe the periulcer mucosal inflammation, the higher the risk of rebleeding. The development of the predictive model for the recurrence of gastric ulcer bleeding after successful treatment based on periulcer mucosal inflammatory status can help clinicians identify high-risk patients early.

[Research on early warning model of the hearing loss of workers exposed to noise]

Objective: To explore the change of hearing threshold of workers exposed to noise, establish an individual-based hearing loss early warning model, accurately and differentiated the health of workers exposed to noise. Methods: In September 2019, all physical examination data of 561 workers exposed to noise from an enterprise were collected since their employment. Three indicators of average hearing threshold of the better ear, namely, at high frequency, 4000 Hz and speech frequency, were constructed. The generalized estimating equation (GEE) was used to adjust gender and age and establish the warning model of each indicator. Finally, sensitive indicators and warning models were screened according to AUC and Yoden index. Results: Among the 561 workers exposed to noise, 26 (4.6%) workers had hearing loss. The sensitivity indicators were the average hearing threshold at speech frequency ≥20 dB, high frequency ≥30 dB and 4000 Hz ≥25 dB. The AUC of each index was 0.602, 0.794 and 0.804, and the Youden indexes were 0.204, 0.588 and 0.608, respectively. In GEE of hearing loss warning models, high-frequency hearing threshold ≥20 dB and 4000 Hz hearing threshold ≥25 dB were the optimal models, with AUC of 0.862. Conclusion: Combined with the changes of individual hearing threshold over the years, can accurately assess the risk of individual hearing loss of workers exposed to noise.

A Novel Early Warning Model for Hand, Foot and Mouth Disease Prediction Based on a Graph Convolutional Network

OBJECTIVES

Hand, foot and mouth disease (HFMD) is a widespread infectious disease that causes a significant disease burden on society. To achieve early intervention and to prevent outbreaks of disease, we propose a novel warning model that can accurately predict the incidence of HFMD.

METHODS

We propose a spatial-temporal graph convolutional network (STGCN) that combines spatial factors for surrounding cities with historical incidence over a certain time period to predict the future occurrence of HFMD in Guangdong and Shandong between 2011 and 2019. The 2011-2018 data served as the training and verification set, while data from 2019 served as the prediction set. Six important parameters were selected and verified in this model and the deviation was displayed by the root mean square error and the mean absolute error.

RESULTS

As the first application using a STGCN for disease forecasting, we succeeded in accurately predicting the incidence of HFMD over a 12-week period at the prefecture level, especially for cities of significant concern.

CONCLUSIONS

This model provides a novel approach for infectious disease prediction and may help health administrative departments implement effective control measures up to 3 months in advance, which may significantly reduce the morbidity associated with HFMD in the future.

Early warning of bloodstream infection in elderly patients with circulating microparticles

Background

The difficulty of early diagnosis of bloodstream infection in the elderly patients leads to high mortality. Therefore, it is essential to determine some new methods of early warning of bloodstream infection in the elderly patients for timely adjustment of treatment and improvement of prognosis.

Methods

Patients aged over 65 years with suspected bloodstream infections were included and divided into bloodstream infection (BSI) and non-bloodstream infection (non-BSI) groups based on blood culture results. The morphology of microparticles (MPs) was observed by using transmission electron microscopy, and the number of MPs was dynamically monitored by flow cytometry.

Results

A total of 140 patients were included in the study: 54 in the BSI group and 86 in the non-BSI group. Total MPs (T-MPs) ≥ 6000 events/µL (OR, 7.693; 95% CI 2.944–20.103, P < 0.0001), neutrophil-derived MPs (NMPs) ≥ 500 events/µL (OR, 12.049; 95% CI 3.574–40.623, P < 0.0001), and monocyte counts ≤ 0.4 × 10⁹/L (OR, 3.637; 95% CI 1.415–9.348, P = 0.007) within 6 h of fever were independently associated with bloodstream infection in the elderly patients. We also developed an early warning model for bloodstream infection in the elderly patients with an area under the curve of 0.884 (95% CI 0.826–0.942, P < 0.0001), sensitivity of 86.8%, specificity of 76.5%, positive predictive value of 70.8%, and negative predictive value of 89.8%.

Conclusion

The early warning model of bloodstream infection based on circulating T-MPs, NMPs, and monocyte counts within 6 h of fever in the elderly patients was helpful in early detection of bloodstream infection and therefore promptly adjustment of treatment plan.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13613-021-00901-w.