Research on the prediction of internet outpatient visits in public hospitals based on ARIMA and GM(1,1)model / 现代医院

Objective To understand the changing trend of Internet outpatient visits in public hospitals,and provide support for the development planning of Internet hospitals.Methods Using the data of Internet outpatient visits in a public hos-pital from January 2021 to June 2023,the ARIMA model and GM(1,1)model were constructed respectively.The mean absolute error(MAE)and root mean square error(RMSE)were used to evaluate the fitting effect,and the Internet outpatient visits from July to December 2023 were predicted based on the dominance model.Results ARIMA(1,2,1)model and GM(1,1)model were used to predict the number of return visits of Internet outpatient service.The average absolute errors were 369.86 and 978.84,and the root-mean-square errors were 479.49 and 1444.83,respectively.The ARIMA(0,1,0)model and GM(1,1)model were used to predict the number of Internet outpatient consultations.The average absolute errors were 297.23 and 369.62,and the root-mean-square errors were 413.61 and 496.30,respectively,indicating that the ARIMA model has a good prediction effect.The forecast results show that the predicted value of Internet outpatient visits in December 2023 is 14,831 cases,and the predicted value of consultation visits is 7461 cases.Conclusion The number of Internet outpatient visits in a public hospital will continue to rise from 2021 to 2023.Therefore,hospitals should fully realize the importance of Internet medical services,take ac-tive measures to continuously optimize the medical service model,and provide patients with high-quality,efficient and convenient Internet medical services.

Predictive analysis of internet-based oncology outpatient visits trend based on ARIMA and GM(1,1)models / 现代医院

Objective To investigate the changing trend of the current situation of Internet-based oncology outpatient treatment and provide support for the development and management of Internet hospitals.Methods The ARIMA and GM(1,1)models were constructed based on the Internet-based outpatient data of a cancer hospital from January 2021 to June 2023,and the fitting effect was evaluated by mean absolute error(MAE)and root mean square error(RMSE).Based on the model,the pro-portion of Internet-based outpatient visits and the offline outpatient visits were predicted from July to December 2023.Results ARIMA(1,1,2)and GM(1,1)models were used to predict the proportion of Internet-based outpatient visits.The average abso-lute errors were 2.06%and 2.41%,and the root-mean-square errors were 3.01%and 3.17%,respectively.The ARIMA(0,1,1)and GM(1,1)models were used to predict the proportion of Internet-based outpatient visits to the offline outpatient visits,with the rate of the average absolute errors of 0.58%and 1.08%,respectively,and the rate of the root mean square errors 0.75%and 1.31%,respectively.The figures indicated that the ARIMA model had a better prediction effect.The forecast results showed that the predicted value of Internet outpatient service in December 2023 was 90.35%,and the predicted value of Internet-based outpatient service accounted for 16.46%of the offline outpatient service.Conclusion In 2021-2023,the proportion of Inter-net-based outpatient visits in the cancer hospital showed a steady trend,and the proportion of Internet outpatient visits in the off-line outpatient visits showed a rising trend.Therefore,hospitals need to establish a continuous monitoring mechanism,constantly adjust management strategies and measures to meet the needs of patients and continue to promote the high-quality development of Internet-based medical services.

Predictive Models for Forecasting Coronavirus Disease 2019 Cases: Relevance to Public Health Services

Introduction@# The global impact of the coronavirus disease 2019 (COVID-19) pandemic has continually jeopardized vulnerable populations encompassing children, youth, elders, and individuals with immunodeficiency and comorbidities. @*Methods@#In recognizing the crucial role of predictive analytics in shaping public health decisions, this study utilizes a predictive design, drawing on official data from the Department of Health (DOH) in the Davao Region, Philippines, spanning 57 days from March 15 to May 10, 2020. By comparing the Susceptible, Infected, Recovered (SIR) model and the Autoregressive Integrated Moving Average (ARIMA) model, the research aims to provide a scientific foundation for informed decision-making by public health authorities. @*Result@#Analysis revealed that the SIR model emerged as the most effective in identifying trends and forecasting future cases. Despite both models indicating a substantial reduction in infection rates, caution is advised against discontinuing control and preventive measures due to the latent potential for another surge. The findings underscore the necessity for scientifically forecasted data to guide decision-makers in enhancing the responsiveness of public health services during similar and potentially worsening conditions. @*Conclusions@#Hence, this study contributes to the ongoing pandemic preparedness and responsiveness discourse. Its emphasis on predictive analytics, particularly the SIR model, offers valuable insights for authorities tasked with safeguarding public health. The significance lies in addressing the current situation in the Davao region and providing a template for future scenarios. As the world grapples with the unpredictable nature of infectious diseases, informed decision-making based on scientific forecasts becomes imperative for effective public health management.

Prevalence status and model analysis of type 2 diabetes death in Urumqi / 公共卫生与预防医学

Objective To analyze the epidemiological characteristics and trends of type 2 diabetes death in Urumqi from 2017 to 2022, and to provide a theoretical basis for formulating diabetes prevention and control policies. Methods The crude mortality rate, standardized mortality rate, annual percentage change (APC), dynamic series and other indicators of 2 177 death data of type 2 diabetes collected in Urumqi from 2017 to 2022 were statistically analyzed. At the same time, the Autoregressive Integrated Moving Average (ARIMA) prediction model was used to establish a model based on the death data from January 1, 2017 to June 30, 2022, to predict the monthly number of type 2 diabetes death in the second half of 2022, and compare it with the actual value to evaluate the model fitting effect. Results From 2017 to 2022, the mortality rate of type 2 diabetes in Urumqi reached 13.46/100 000, and the standardized mortality rate was 11.78/100 000. There was a significant difference in APC results for male mortality (P<0.05). The mortality rate of type 2 diabetes increased with age, and the mortality rate was higher in men than in women before the age of 70, and conversely, female mortality was higher than male. Retirees, married people, and people with junior high school education or below had higher mortality rates than others. The results of the standardized mortality dynamic series showed that the average rate of development in men was higher than that in the general population and women. By establishing the optimal ARIMA (0,1,1) prediction model, the model fit was qualified, while the accuracy would need to be improved. Conclusion From 2017 to 2022, the mortality rate of type 2 diabetes in Urumqi has an increasing trend. In order to realize the Healthy China Action, it is necessary to focus on health publicity and education for the elderly in the jurisdiction, prevent the occurrence of type 2 diabetes comorbidities, and reduce the mortality rate of type 2 diabetes.

Time characteristics and trend prediction of hepatitis B infection in Baqiao Distract,Xi'an City / 公共卫生与预防医学

Objective To analyze the characteristics of hepatitis B infection time in Baqiao Distract of Xian City, predict the incidence trend of hepatitis B in the next years with ARIMA model, and provide theoretical guidance for the adjustment of hepatitis B prevention and control strategies. Methods A total of 7173 cases of hepatitis B in our hospital from 2018 to 2023 were collected as subjects,the annual percentage of change (APC) was used to analyze the characteristics of time of hepatitis B infection in the next years, and the ARIMA optimal model was used to predict the incidence rate in 2024. Results The gross incidence of hepatitis B in Baqiao Distract of Xian City from 2018 to 2023 was 51.30 per 100 000, and the standardized incidence rate was 42.11 per 100 000. The six-year incidence rate showed an upward trend (APC=8.71%,95% CI：3.29% -9.61% , P<0.05). Chi square results showed that the standardized incidence rate of hepatitis B in men is 1.51 times than women, and the incidence rate of hepatitis B in the 30-45 age group was the highest. The prediction results showed that the number of hepatitis B cases in 2024 was 1590, which was 1.91% lower than that in 2023 (1621 cases). Conclusion The infection of hepatitis B is on the rise in Baqiao Distract of Xian City, ARIMA can be used to predict this kind of infectious disease and provide scientific guidance for the prevention of this disease.

Application of GM(1,1) model modified with seasonal factors in prediction of PM2.5 concentration in Shanghai / 公共卫生与预防医学

Objective To explore PM2.5 concentration modeling and prediction based on the monthly average concentrations of PM2.5 in Shanghai since 2015, and to provide new ideas about PM2.5 prediction methods. Methods The seasonal factors were introduced into the Grey Model (GM). GM(1,1) model modified with seasonal factors was established and compared with seasonal autoregressive integrated moving average model (ARIMA) model. The data of 2015-2021 was used for modeling and prediction, and the data from January to October in 2022 was used as a validation set to evaluate the prediction effectiveness. The monthly average PM2.5 concentrations in Shanghai from November to December in 2022 were predicted. Results Seasonal ARIMA model showed RMSE=4.02 and MAPE=15.50% in the validation set, while GM(1,1) model modified with seasonal factors showed RMSE=3.30 and MAPE=11.59%. GM(1,1) model modified with seasonal factors predicted the monthly average PM2.5 concentrations in Shanghai from November to December in 2022 to be 24.99 and 34.83μg/m3, respectively. Conclusion The prediction effect of GM(1,1) model modified with seasonal factors has better predictive performance than seasonal ARIMA model. The grey prediction model modified with seasonal factors can be considered when predicting seasonal time series such as the concentration of PM2.5.

Predictive analysis of the number of hospitalized patients with acute pancreatitis based on time series model / 中华胰腺病杂志

Objective:To predict and analyze the number of acute pancreatitis (AP) inpatients based on time series model, and to explore the predictive efficiency of the model.Methods:Clinical data of AP inpatients in the Affiliated Hospital of Southwest Medical University from January 2014 to December 2019 were collected. R software was used to collect the time series of AP inpatients, and the trend and seasonal characteristics of AP inpatients from 2014 to 2018 were analyzed. Furthermore, the autoregressive moving average (ARIMA) model was established through stationarity test, model ordering and model testing steps, and the best selected model was used to predict the monthly number of inpatients in 2019 to verify its prediction efficiency.Results:A total of 3 939 AP patients were included in the study. The most common etiology for AP was cholestrogenic (48.2%), followed by hyperacylglyceremia (36.3%). The peak age of hospitalization was from 40 to 60 years old. Time series analysis showed that the number of AP inpatients increased year by year. The highest peak of the disease was from February to March, followed by September to November; and there was seasonal variation and the incidence was relatively small in summer. The established original training set sequence did not pass the stationarity test ( P=0.061), so the ARIMA model was established after it was transformed into a stationarity sequence by first-order difference. According to the criterion of minimum AIC value, ARIMA(2, 1, 1)(1, 1, 1) 12 was selected as the best model. The model was used to predict the number of AP inpatients in 2019, showing that it could better fit the trend of onset time and had good short-term prediction effect. The mean root error and absolute error were 6.8790 and 4.7783, respectively. Conclusions:The number of AP inpatients increases year by year with seasonal changes. ARIMA model is effective in predicting the number of AP inpatients and can be used for short-term prediction.

Time series prediction analysis of average hospitalization expenditure and drug proportion of diabetes patients in a grade-a tertiary hospital in Zhejiang province / 现代医院

Objective This study conducted the ARIMA model to analyze the cost structure and trend of inpatients with diabetes in a grade-A tertiary hospital and provide scientific basis for effectively controlling diabetes hospitalization expenses and reduce patient's economic burden.Methods The data of 18 371 inpatients with diabetes from 2012 to 2022 in a grade-A tertiary hospital were collected.We collected inpatient data of diabetes from 2012 to 2021 to fit the average inpatient expenses and drug proportion,and used data 2022 to verify the effect of model prediction.We predicted the average inpatient expenses and drug proportion from 2023 to 2025.Results The difference between the predicted value and the actual value was small,and the mean absolute percentage error was within the acceptable range.ARIAM model could be used to predict the expenses of diabetes.Conclusion The average cost of hospitalization showed a decreasing trend,and"the five colors,one map,one index"manage-ment model has achieved results.The proportion of drugs decreased obviously,but the composition of hospitalization expenses should be further optimized.The research of expenses prediction based on diabetes needs to be depended.

Epidemiological characteristics and prediction of varicella trends in Yangpu District， Shanghai， from 2005 to 2022 / 上海预防医学

ObjectiveTo analyze the epidemiological characteristics of varicella in Yangpu District， Shanghai from 2005 to 2022， predict the trend of varicella in Yangpu District in 2023， and provide evidence for prevention and control of varicella outbreaks. MethodsInformation of varicella cases reported in Yangpu District from 2005 to 2022 was obtained from the China Information System for Disease Control and Prevention. Descriptive statistics was used to characterize the varicella epidemiology. An autoregressive integrated moving average （ARIMA） model was established by using the number of cases per month from 2005 to2022 to predict the trend of varicella epidemics in Yangpu District in 2023. The varicella incidence in 2022 was used to evaluate the fitness of the ARIMA model. ResultsFrom January 2005 to December 2021， a total of 11 527 cases of varicella were reported in Yangpu District， Shanghai. After excluding duplicates and clinical diagnoses， 11 413 cases were included into the analysis. The annual average incidence rate was 51.87/105， the age of onset was mainly under 20 years old （66.5%）， and the occupation was mainly students （49.7%）. The ARIMA （1，1，0）×（0，1，1）12 model was constructed and showed a good fitness while using monthly reported varicella cases in 2022 for model fitting. It was predicted that 1 089 cases of varicella would be reported in Yangpu District in 2023. ConclusionIt is predicted that varicella cases in Yangpu District will increase in 2023. It is recommended to continue promoting delayed varicella vaccination to maintain a high level of vaccination rate. Before the peak of the epidemic， health education regarding varicella should be strengthened， and measures for epidemic prevention and control should be reinforced to prevent varicella outbreaks.

Application of ARIMA model in predicting pediatric red blood cell usage / 中国输血杂志

【Objective】 To explore the feasibility of using autoregressive moving average model (ARIMA) to predict the dosage of suspended red blood cells in children, and to provide a basis for the development of clinical blood reserve plans in children's hospitals. 【Methods】 ARIMA model was constructed using the total blood consumption of clinical suspended red blood cells from March 2016 to May 2022 at the Children's Hospital of Chongqing Medical University as the data source by SPSS26.0 software. The optimal model was used to predict the clinical suspended red blood cell consumption from June to October 2022, and the predictive effect of the model was tested. 【Results】 ARIMA(0, 1, 1) (0, 1, 1)12 was the optimal model for predicting the consumption of suspended red blood cells in pediatrics. The autocorrelation function and partial autocorrelation function of the residual sequence basically fell within the 95% confidence interval. At the same time, Ljung-Box Q statistical results showed that there was no correlation between the residual (P>0.05), indicating that the residual was white noise, which met the randomicity hypothesis. The average relative error between the predicted values of the model and the actual clinical red blood cell usage from June to October 2022 was 5%, indicating high prediction accuracy. 【Conclusion】 The blood usage of children has obvious seasonal and periodic patterns, and the optimal model ARIMA (0, 1, 1) (0, 1, 1)12 can better fit the trend of changes in pediatric suspended red blood cell usage, thus providing a basis for the development of clinical blood reserve plans in children's hospitals.

The predictive effect of ARIMA model for occupational pneumoconiosis in Guangdong Province / 中国职业医学

Objective: To verify the accuracy of the autoregressive integrated moving average (ARIMA) in predicting the incidence of occupational pneumoconiosis (hereinafter referred as pneumoconiosis) and to predict the incidence of pneumoconiosis in Guangdong Province in the next five years. Methods: A follow-up survey was performed to collect data on pneumoconiosis patients reported in Guangdong Province from 1956 to 2021. Collected data from 1956 to 2016 were used as the training set to build an ARIMA model. Collected data from 2017 to 2021 were used as the prediction set to evaluate the predicting result of the ARIMA model. The ARIMA model was used to predict the incidence of pneumoconiosis in Guangdong Province in next five years. Results: The ARIMA (1,1,2) model was set up after model identification and order estimation. The model was used to predict the prediction set, and its result was good. The ARIMA result and actual values in 2021 were 213 and 210 cases, respectively, with a difference of only three cases. The number of pneumoconiosis cases predicted using the ARIMA model in Guangdong Province from 2022 to 2026 was 214, 204, 202, 194, and 191 cases, respectively, showing a trend of low-level prevalence. Conclusion: The ARIMA model demonstrates high accuracy in predicting pneumoconiosis incidence over a long period of time and with large sample sizes. The forecast results of the ARIMA(1,1,2) model indicate that the incidence of pneumoconiosis in Guangdong Province will be around 200 cases in the next five years, indicating a low-level prevalence.

Epidemic status and model analysis of cardiovascular and cerebrovascular diseases attributed to renal insufficiency in elderly Chinese / 公共卫生与预防医学

Objective To analyze the changing trend of disease burden attributable to renal insufficiency in cardiovascular disease (CVD) among the elderly in China from 1990 to 2019, and to forecast the disability-adjusted life years (DALY) in the next 10 years, so as to provide a reference basis for accurate prevention and control of CVD attributable to renal insufficiency in China. Methods Data were obtained from the Global Health Data Exchange (GHDx) database to describe the current status of CVD prevalence attributable to renal insufficiency. The joinpoint model was used to estimate the annual percentage change and average annual percentage change to assess the temporal trend of CVD attributable to renal insufficiency in China. An autoregressive moving average model was created by R4.0.2 software to predict the disease burden of CVD attributable to renal insufficiency in China. Results Compared with 1990, CVD mortality and DALY rates attributed to renal insufficiency increased in the male elderly population and decreased in women. Mortality and DALY rates attributed to ischemic heart disease, ischemic stroke, and peripheral arterial disease attributed to renal insufficiency showed an increasing trend, and mortality and DALY rates for cerebral hemorrhage decreased. There was an overall increasing trend in the attribution of CVD due to renal insufficiency. Conclusion The burden of diseases attributable to renal insufficiency in Chinese elderly with CVD is relatively high, and the impact on each disease is different, which requires the attention of relevant authorities.

ARIMA model in predicting of COVID-19 epidemic for the Southern Africa region

Background: Coronavirus pandemic, a serious global public health threat, affects the Southern African countries more than any other country on the continent. The region has become the epicenter of the coronavirus with South Africa accounting for the most cases. To cap the deadly effect caused by the pandemic, we apply a statistical modelling approach to investigate and predict COVID-19 incidence. Methods: Using secondary data on the daily confirmed COVID-19 cases per million for Southern Africa Development Community (SADC) member states from March 5, 2020, to July 15, 2021, we model and forecast the spread of coronavirus in the region. We select the best ARIMA model based on the log-likelihood, AIC, and BIC of the fitted models. Results: The ARIMA (11,1,11) model for the complete data set was finally selected among ARIMA models based upon the parameter test and the Box­Ljung test. The ARIMA (11,1,9) was the best candidate for the training set. A 15-day forecast was also made from the model, which shows a perfect fit with the testing set. Conclusion: The number of new COVID-19 cases per million for the SADC shows a downward trend, but the trend is characterized by peaks from time to time. Tightening up of the preventive measures continuously needs to be adapted in order to eradicate the coronavirus epidemic from the population.

Comparison of random forests and SARIMA in Predicting Brucellosis Incidence / 公共卫生与预防医学

Objective To compare the effects of random forest and SARIMA (Seasonal Autoregressive Integrated Moving Average) on predicting incidence rate of brucellosis. Methods Using Brucellosis cases reported in the China Disease Prevention and Control Information System from 2005 to 2017, two models, random forest and SARIMA, were established for training and forecasting, and the forecasting results of the two models were compared. Results The R2 (R Squared) and RMSE (Root Mean Squared Error) of SARIMA model and random forest model are 0.904, 0.034351, 0.927 and 0.03345 respectively. Conclusion Both models have high prediction accuracy and can predict the incidence of brucellosis. Random forest prediction is a little bit better than SARIMA model and has more practical value.

Epidemiological characteristics analysis and incidence trend prediction of gonorrhea in Hubei Province in 2010-2021 / 公共卫生与预防医学

Objective To analyze the epidemiological characteristics and incidence trend of gonorrhea in Hubei Province, and to provide reference for scientific formulation of prevention and control measures. Methods Based on the surveillance data of gonorrhea from 2010 to 2021, three-way distribution and ARIMA model were used for data analysis and incidence prediction. Results From 2010 to 2021, the reported incidence rate fluctuated between 3.01/100 000-7.07/100 000, and the average annual reported incidence rate was 4.62/100 000. The reported incidence rate showed the characteristics of “first fall and then rise, and then fall and rise again”, and the peak incidence period was from June to December. The male to female ratio of reported cases was 5.78:1, and the number of reported cases in the age group of 20-39 years old accounted for 62.43% of the total number of cases. The reported cases were mainly housework and unemployed, farmers, and unknown occupation. The severity of the regional incidence was divided into 5 categories by the Q-type clustering, and the most serious category included Shennongjia Forest District, Huangshi City, and Wuhan City. The ARIMA model predicted the incidence rate to be in good agreement with the actual incidence rate, with a predicted number of 3 343 cases in 2022. Conclusion At present, gonorrhea in Hubei Province is still at a high prevalence level. There are obvious differences in gender, age, occupation, and regional distribution. The ARIMA model is suitable for predicting the incidence of gonorrhea, and it is predicted that the incidence will increase slightly in 2022.

Prediction model of the number of street voluntary blood donors / 中国输血杂志

【Objective】 To explore the relationship between climate factors and the number of street voluntary blood donors in Beijing and develop a reliable predictive model, so as to provide reference for donor recruitment. 【Methods】 The data of weather and the number of street blood donors from January 2018 to October 2019 were collected to formulate generalized additive model(GAM) and autoregressive integrated moving average model(ARIMA), and the predicative accuracy of the two models was assessed using data from November to December 2019. 【Results】 GAM indicated that the number of donors decreased when the wind force was 4 to 5 (95%CI: 0.805, 0.995), and the number on weekends and official holidays was 1.562 (95% CI: 1.510, 1.617) and 1.779 (95%CI: 1.035, 3.055) times that of the working day respectively. The number of blood donors increased with the elevation of temperature until 25℃, then declined with temperature increasing slowly. The two-day predictive accuracy of GAM and ARIMA was 92.14% and 90.55%, with overall accuracy at (84.46±11.12)% and (87.65±9.3)%, respectively. 【Conclusion】 Considering official holiday, strong wind and temperature, etc, the ARIMA model runs stable overall, while GAM is good at short-term prediction. The comprehensive use of two predictive models is helpful in guiding the recruitment of blood donors.

Study on prediction of clinical demand for plasma components in Suzhou city based on ARIMA model / 中国输血杂志

【Objective】 To establish a prediction model of clinical blood demand in Suzhou urban area by ARIMA model, and to predict future clinical blood demand by sorting out the historical data, so as to guide the reasonable collection and scientific deployment of blood resources, and achieve the balance of clinical blood supply and demand. 【Methods】 The monthly data of clinical use of plasma components in Suzhou city from 2009 to 2019 were obtained, and analyzed by SPSS26 software and ARIMA model. Through model identification, parameter estimation and optimal model test, the optimal model for clinical blood prediction was determined and used to predict the clinical consumption of plasma components from January to November 2020. The predicted value was compared with the actual value to verify the prediction effect of the model. 【Results】 The optimal model was ARIMA(0, 1, 1)(0, 1, 1)12. The values of ACF autocorrelation function and PACF partial autocorrelation function of residual were both within 95%CI. Meanwhile, the Yang-Box Q statistic value was 11.596, P>0.05, which passed the white noise test. The predicted values of clinical consumption of plasma components in Suzhou urban area from January to November 2020 were all within 95%CI, consistent with the trend of actual values, with small mean relative error(7.9%) and good prediction effect. 【Conclusion】 ARIMA model can be used for short-term prediction on clinical use of plasma components in Suzhou city, and provide reference for reasonable collection, preparation and scientific deployment.

Prediction on clinical platelet demand in Suzhou based on ARIMA model / 中国输血杂志

【Objective】 To establish an ARIMA model suitable for clinical platelet demand prediction in Suzhou, which can be used as reference to predict future clinical platelet demand and provide scientific basis for platelet collection, preparation, stock management and clinical deployment for blood banks, so as to achieve the maximum balance between platelets supply and demand . 【Methods】 The data of platelet consumption in Suzhou from 2009 to 2019 were collected and analyzed by SPSS 26 software, Time series analysis method was used to establish the ARIMA model. The model was further optimized through model identification, parameter estimation and optimal model test, and then used to predict clinical platelet consumption from January to November 2020. The predicted value was compared with the actual value to verify the prediction effect of the model. 【Results】 The optimal model for the prediction of platelet clinical demand was ARIMA (0, 1, 1) (0, 1, 1) 12. The ACF autocorrelation function value and PACF partial autocorrelation function value of the residuals were within 95% CI. Meanwhile, the LJUNG BOX test was 13.982 (P>0.05), indicating that there was no autocorrelation in the residuals. The trend of the curve between the predicted and actual value was basically the same(except for February 2020), and the predicted values were within 95% CI, with the average relative error of 7.22%, which was lower than 10%, showing good prediction effect. 【Conclusion】 ARIMA model can be used for short-term prediction of clinical platelets demand in Suzhou, and can provide basis for reasonable collection, preparation and deployment of platelets.

ARIMA model in blood supply prediction of multi blood components / 中国输血杂志

【Objective】 To establish an ARIMA model to fit the distributed units of four blood components from 2010 to 2018 in Tianjin and test the fitting degree, so as to predict the future issuing units of these blood products, and provide scientific basis for the blood center to formulate blood collection and donor recruitment plan. 【Methods】 The monthly distributed data of blood components from 2010 to 2019 were sorted out to establish the ARIMA model. The model identification, parameter estimation and test of the distributed data concerning red cells, plasma, apheresis platelet and white cells from January 2010 to December 2019 were performed to determine the optimal model using Eviews 10.0 software. Considering the obvious trend and seasonality of data, the seasonal model was chosen to predict the issuing of four blood products in January to December 2019, and the fitting degree was tested by comparing with the actual value. 【Results】 The ARIMA model residual autocorrelation function and partial autocorrelation function of four blood components showed that the regression residuals of each product had the same variance. The predicted value of supply was basically within 95% CI, and the curve trend of model fitting value and actual value was basically consistent, The average relative errors of red cells, plasma, apheresis platelets and white cells were 6.19%, 5.08%, 1.72% and 7.17%, respectively. 【Conclusion】 ARIMA model can appropriately fit the change trend of blood supply in Tianjin, which is helpful to understand the clinical requirements in the near future, provide the basis for blood collection, recruitment and inventory management.

A comparative study of ARIMA and LSTM models in predicting hospital discharge number / 公共卫生与预防医学

Objective To fit and predict the monthly discharge number of a specialist hospital using Autoregressive Integrated Moving Average model (ARIMA) and Long Short-Term Memory Neural Network model (LSTM), and compare the prediction effects of the two models. Methods ARIMA and LSTM models were constructed based on the monthly discharge number of a specialist hospital from 2013 to 2018. The resulting models were then used to predict the monthly discharge numbers in 2019, which were compared with actual data. The mean absolute percentage error (MAPE) was used to evaluate the prediction effect of these two models. Results The MAPE values of ARIMA and LSTM compared to actual data in 2019 were 7.90% and 14.26%, respectively. Conclusion The prediction effect of ARIMA was better than that of LSTM. The prediction results of ARIMA showed that the number of patients discharged from the specialist hospital in 2019 was increasing, which fit well with the actual data.