| Objective1.To analyze the epidemiological characteristics of imported malaria in Jiangsu Province from 2012 to 2019.2.To analyze the current situation of care-seeking delay of imported malaria in Jiangsu province.3.To analyze the influencing factors of care-seeking delay of imported malaria in Jiangsu Province.4.To construct the risk prediction models of care-seeking delay of imported malaria patients in Jiangsu Province.5.To develop a decision support system for imported malaria patients for monitoring imported malaria and predicting the risk of delay in seeking medical care after symptom onset.MethodsBased on the information management system for reporting infectious diseases and the information management system for parasitic disease prevention and control of the CISDCP,information on case investigation,first symptoms and time of initial diagnosis of imported malaria cases reported in Jiangsu Province from 2012 to 2019 was collected.The epidemiological analysis was conducted to describe the epidemiological status of imported malaria in Jiangsu Province according to the "three-dimensional distribution",and to calculate constituent ratio of different years.The frequency and percentage of imported malaria patients were used for the analysis of the categorical data of medical treatment.Univariate analysis was performed based on two conditions and compared using Chi-square test,continuity correction or Fisher’s exact probability method with a test level of α=0.05.After dividing the data into training and validation sets according to 7:3,five different models for predicting the risk of delay in seeking medical care for imported malaria cases were constructed using BP neural network,logistic regression,LSVM,random forest,and Bayesian respectively and to validate the models.Logistic regression was analyzed visually using nomogram,calibration curves were drawn to evaluate the column line plots,and the AUC of the five models were compared to evaluate their predictive efficacy.Then,a prediction system for the risk of care-seeking delay for imported malaria patients in Jiangsu province was constructed based on the logistic regression model.Results1.Epidemiological characteristics of imported malaria in Jiangsu provinceA total of 2 255 patients with imported malaria in Jiangsu Province were included from 2012 to 2019,among which the majority were male young adults.The number of imported malaria cases in Jiangsu Province showed an increasing trend from 2012 to 2015,decreased significantly from 2016 to 2017,and was more stable in 2018 and 2019.The peak incidence is usually during the winter return period,and the top three cities in terms of incidence are Nantong,Lianyungang and Taizhou.The infected Plasmodium species is mainly P.falciparum.2.The current status of care-seeking delay of imported malaria in Jiangsu ProvinceAmong the malaria cases reported in Jiangsu Province from 2012 to 2019,71.80%(1 619/2 255)were seen 24 hours after the onset of first symptoms,of which 43.37%(978/2 255)were seen at 72 hours and above,and only 28.20%(636/2 255)were seen on the same day of the onset of first symptoms.Initial consultations for malaria cases were concentrated in county and municipal healthcare institutions,with fewer visits to township healthcare institutions,village clinic,and provincial healthcare institutions.The results of univariate analysis using whether or not the delay in seeking medical treatment was defined beyond 24 hours as the dependent variable showed statistically significant differences(P<0.05)in month of return,history of malaria infection,time of onset of symptoms,and level of health facilities for initial healthcare;the results of univariate analysis using whether or not the delay in seeking medical treatment was defined beyond 72 hours as the dependent variable showed that month of return,history of malaria infection,time between last and current infection.The differences were statistically significant(P<0.05)for month of return,history of malaria infection,time between last and current infection,time to onset of symptoms,level of first consultation unit,country of origin,and GDP of the reporting city.3.Influencing Factors in care-seeking delay of imported malaria in Jiangsu ProvinceA binary logistic regression analysis was performed using the delay in seeking medical care after 24 hours after the onset of first symptoms as the delay in seeking medical care,and the results showed that the onset of symptoms within one week to one month after entry(P=0.048,OR=0.116,95%CI:0.014～0.980)and level of health facilities for initial healthcare were included in the regression model,and level of health facilities for initial healthcare included township healthcare institutions(P<0.001.OR=2.976,95%CI:1.777～4.953),county healthcare institutions(P<0.001,OR=6.491,95%CI:4.338～9.713),municipal healthcare institutions(P<0.001,OR=6.432,95%CI:4.173～9.912),provincial healthcare institutions and others(P<0.001,OR=5.070,95%CI:2.479～10.367),A total of 3 factors were included in the regression model for the analysis of the delay in seeking care by seeking care after 72 hours after the onset of first symptoms,namely reporting city GDP(2)(P=0.007,OR=0.634,95%CI:0.457～0.881),time to onset of symptoms,and level of health facilities for initial healthcare,with time to onset of symptoms including one week after entry(P=0.003,OR=0.183,95%CI:0.060～0.563),one week to one month after entry(P<0.001,OR=0.122,95%CI:0.040～0.375),and one month after entry(P=0.016,OR=0.240,95%CI:0.075～0.763).Level of health facilities for initial healthcare included township healthcare institutions(P=0.019,OR=2.401,95%CI:1.155～4.991),county healthcare institutions(P<0.001,OR=3.833,95%CI:2.096～7.006),municipal healthcare institutions(P<0.001,OR=4.103,95%CI:2.207～7.627),provincial healthcare institutions and others(P<0.001,OR=5.702,95%CI:2.512～12.945).4.Predicting the risk of care-seeking delay of imported malaria patients in Jiangsu ProvinceWe constructed and compared five machine learning models based on machine learning algorithms:BP neural network,LSVM,Bayesian,logistic regression and random forest,and found that the BP neural network model had the best prediction performance.The BP neural network model was constructed with the delay in seeking medical treatment after 24 hours from the symptom onset.The results showed that the five factors that had a greater impact on the delay in seeking medical treatment for imported malaria were the level of health facilities for initial healthcare,the time of symptom onset,the time interval between the last and the current infection,the month of return and the symptoms.The BP neural network model was constructed with the delay in seeking medical treatment after 72 hours after the symptom onset,and the results showed that the five factors that had a greater influence on the delay in seeking medical treatment for imported malaria were the time of symptom appearance,the level of health facilities for initial healthcare,the country of origin,the symptom and the month of return.Conclusion1.Imported malaria cases in Jiangsu province are mainly male young adults,and the peak incidence is usually in the winter homecoming period,and the infected Plasmodium falciparum species are mainly Plasmodium falciparum.This part of the results suggests that we should focus on the prevention of malaria among foreign workers.Health education for this population should be strengthened to raise awareness of early medical consultation and reduce the occurrence of serious illness and death.Strengthen monitoring efforts during the peak homecoming period to prevent re-transmission of imported malaria.2.There is a delay in care-seeking delay of imported malaria cases in Jiangsu Province,and the initial diagnosis of malaria cases is mainly concentrated in county and municipal healthcare institutions.This suggests that the awareness of timely medical treatment for imported malaria patients in China needs to be improved.3.The probability of care-seeking delay after 24 hours after the first symptom appeared was higher for imported malaria patients with symptoms before entry and higher level of initial diagnosis units.The higher the GDP level of the reporting city,the higher the probability that imported malaria patients with pre-entry symptoms and higher the level of health facilities for initial healthcare would seek medical treatment after 72 hours of the symptom onset.This suggests the need for focused screening and surveillance of this category of entrants to prevent local transmission from imported malaria cases from abroad.There is also a need to strengthen the dissemination of malaria control knowledge in our primary care institutions.4.The BP neural network model had the best predictive efficacy compared to the four machine learning models LSVM,Bayesian,logistic regression,and random forest,followed by the logistic regression model. |
PDF Full Text Request