Predicting 1-year mortality of patients with diabetes mellitus in Kazakhstan based on administrative health data using machine learning.

Diabetes mellitus (DM) affects the quality of life and leads to disability, high morbidity, and premature mortality. DM is a risk factor for cardiovascular, neurological, and renal diseases, and places a major burden on healthcare systems globally. Predicting the one-year mortality of patients with DM can considerably help clinicians tailor treatments to patients at risk. In this study, we aimed to show the feasibility of predicting the one-year mortality of DM patients based on administrative health data. We use clinical data for 472,950 patients that were admitted to hospitals across Kazakhstan between mid-2014 to December 2019 and were diagnosed with DM. The data was divided into four yearly-specific cohorts (2016-, 2017-, 2018-, and 2019-cohorts) to predict mortality within a specific year based on clinical and demographic information collected up to the end of the preceding year. We then develop a comprehensive machine learning platform to construct a predictive model of one-year mortality for each year-specific cohort. In particular, the study implements and compares the performance of nine classification rules for predicting the one-year mortality of DM patients. The results show that gradient-boosting ensemble learning methods perform better than other algorithms across all year-specific cohorts while achieving an area under the curve (AUC) between 0.78 and 0.80 on independent test sets. The feature importance analysis conducted by calculating SHAP (SHapley Additive exPlanations) values shows that age, duration of diabetes, hypertension, and sex are the top four most important features for predicting one-year mortality. In conclusion, the results show that it is possible to use machine learning to build accurate predictive models of one-year mortality for DM patients based on administrative health data. In the future, integrating this information with laboratory data or patients' medical history could potentially boost the performance of the predictive models.

Review of the research databases on population-based Registries of Unified electronic Healthcare system of Kazakhstan (UNEHS): Possibilities and limitations for epidemiological research and Real-World Evidence.

INTRODUCTION: A 'learning healthcare system', based on electronic health records and other routinely collected healthcare data, would allow Real World Data (RWD) to be continuously fed into the system, ensuring that with every new patient treated, we know more overall about the practice of medicine. A judicious use of RWD would complement the traditional evidence from clinical research, for the benefit of all stakeholders involved in healthcare. Lack of data on disease epidemiology in Kazakhstan resonates with lower life expectancy and poorer health indicators compared to countries with analogous income per capita. Usage of primary data collection methods to fill these gaps require additional financial and human resources. Usage of big data, which is routinely collected though healthcare information systems, is considered as a competitive alternative in described circumstances. OBJECTIVE: Development of the Unified National Electronic Healthcare System (UNEHS) in Kazakhstan allowed the creation of research databases to investigate epidemiology of numerous diseases. UNEHS research databases endorse extensive research activities due to a prospective follow-up, coverage of the whole Kazakhstani population and relatively lower expenses to conduct epidemiological studies. This review paper aims to introduce the content and descriptive data on research databases on population-based registries of UNEHS and to discuss opportunities and limitations of its usage. RESULTS AND DISCUSSION: UNEHS databases include medical data on 36.4% of an adult population of Kazakhstan. Research databases presented in this paper contain critical variables that can be utilized for investigation of disease epidemiology, effectiveness of provided medical procedures and infectious disease epidemiology. A few examples accompany a detailed elaboration on the possibilities of research database utilization in epidemiological research. CONCLUSION: Considering numerous advantages, the UNEHS research databases are expected to greatly contribute to healthcare in Kazakhstan by providing critical data on disease epidemiology. To warrant long-term usage and high research output several concerns and limitations should be addressed as well.

Epidemiological characteristics and climatic variability of viral meningitis in Kazakhstan, 2014-2019.

Background: The comprehensive epidemiology and impact of climate on viral meningitis (VM) in Kazakhstan are unknown. We aimed to study the incidence, in-hospital mortality and influence of climatic indicators on VM from 2014 to 2019. Methods: Nationwide electronic healthcare records were used to explore this study. ICD-10 codes of VM, demographics, and hospital outcomes were evaluated using descriptive statistics and survival analysis. Results: During the 2014-2019 period, 10,251 patients with VM were admitted to the hospital. 51.35% of them were children, 57.85% were males, and 85.9% were from the urban population. Enteroviral meningitis was the main cause of VM in children. The incidence rate was 13 and 18 cases per 100,000 population in 2014 and 2019, respectively. Case fatality rate was higher in 2015 (2.3%) and 2017 (2.0%). The regression model showed 1°C increment in the daily average temperature might be associated with a 1.05-fold (95% CI 1.047-1.051) increase in the daily rate of VM cases, 1hPa increment in the average air pressure and 1% increment in the daily average humidity might contribute to a decrease in the daily rate of VM cases with IRRs of 0.997 (95% CI 0.995-0.998) and 0.982 (95% CI 0.981-0.983), respectively. In-hospital mortality was 35% higher in males compared to females. Patients residing in rural locations had a 2-fold higher risk of in-hospital death, compared to city residents. Elderly patients had a 14-fold higher risk of in-hospital mortality, compared to younger patients. Conclusion: This is the first study in Kazakhstan investigating the epidemiology and impact of climate on VM using nationwide healthcare data. There was a tendency to decrease the incidence with outbreaks every 5 years, and mortality rates were higher for Russians and other ethnicities compared to Kazakhs, for males compared to females, for elder patients compared to younger patients, and for patients living in rural areas compared to city residents. The climatic parameters and the days of delay indicated a moderate interaction with the VM cases.