Dementia risk predictions from German claims data using methods of machine learning.

INTRODUCTION: We examined whether German claims data are suitable for dementia risk prediction, how machine learning (ML) compares to classical regression, and what the important predictors for dementia risk are. METHODS: We analyzed data from the largest German health insurance company, including 117,895 dementia-free people age 65+. Follow-up was 10 years. Predictors were: 23 age-related diseases, 212 medical prescriptions, 87 surgery codes, as well as age and sex. Statistical methods included logistic regression (LR), gradient boosting (GBM), and random forests (RFs). RESULTS: Discriminatory power was moderate for LR (C-statistic = 0.714; 95% confidence interval [CI] = 0.708-0.720) and GBM (C-statistic = 0.707; 95% CI  = 0.700-0.713) and lower for RF (C-statistic = 0.636; 95% CI  = 0.628-0.643). GBM had the best model calibration. We identified antipsychotic medications and cerebrovascular disease but also a less-established specific antibacterial medical prescription as important predictors. DISCUSSION: Our models from German claims data have acceptable accuracy and may provide cost-effective decision support for early dementia screening.

Diabetes duration and the risk of dementia: a cohort study based on German health claims data.

OBJECTIVE: Diabetes is a risk factor for dementia but little is known about the impact of diabetes duration on the risk of dementia. We investigated the effect of type 2 diabetes duration on the risk of dementia. DESIGN: Prospective cohort study using health claims data representative for the older German population. The data contain information about diagnoses and medical prescriptions from the in- and outpatient sector. METHODS: We performed piecewise exponential models with a linear and a quadratic term for time since first type 2 diabetes diagnosis to predict the dementia risk in a sample of 13,761 subjects (2,558 dementia cases) older than 65 years. We controlled for severity of diabetes using the Adopted Diabetes Complications Severity Index. RESULTS: We found a U-shaped dementia risk over time. After type 2 diabetes diagnosis the dementia risk decreased (26% after 1 year) and reached a minimum at 4.75 years, followed by an increase through the end of follow-up. The pattern was consistent over different treatment groups, with the strongest U-shape for insulin treatment and for those with diabetes complications at the time of diabetes diagnosis. CONCLUSIONS: We identified a non-linear association of type 2 diabetes duration and the risk of dementia. Physicians should closely monitor cognitive function in diabetic patients beyond the first few years after diagnosis, because the later increase in dementia occurred in all treatment groups.

Prediction-based variable selection for component-wise gradient boosting.

Model-based component-wise gradient boosting is a popular tool for data-driven variable selection. In order to improve its prediction and selection qualities even further, several modifications of the original algorithm have been developed, that mainly focus on different stopping criteria, leaving the actual variable selection mechanism untouched. We investigate different prediction-based mechanisms for the variable selection step in model-based component-wise gradient boosting. These approaches include Akaikes Information Criterion (AIC) as well as a selection rule relying on the component-wise test error computed via cross-validation. We implemented the AIC and cross-validation routines for Generalized Linear Models and evaluated them regarding their variable selection properties and predictive performance. An extensive simulation study revealed improved selection properties whereas the prediction error could be lowered in a real world application with age-standardized COVID-19 incidence rates.

Social disparities in the first wave of COVID-19 incidence rates in Germany: a county-scale explainable machine learning approach.

OBJECTIVES: Knowledge about the socioeconomic spread of the first wave of COVID-19 infections in Germany is scattered across different studies. We explored whether COVID-19 incidence rates differed between counties according to their socioeconomic characteristics using a wide range of indicators. DATA AND METHOD: We used data from the Robert Koch-Institute (RKI) on 204 217 COVID-19 diagnoses in the total German population of 83.1 million, distinguishing five distinct periods between 1 January and 23 July 2020. For each period, we calculated age-standardised incidence rates of COVID-19 diagnoses on the county level and characterised the counties by 166 macro variables. We trained gradient boosting models to predict the age-standardised incidence rates with the macrostructures of the counties and used SHapley Additive exPlanations (SHAP) values to characterise the 20 most prominent features in terms of negative/positive correlations with the outcome variable. RESULTS: The first COVID-19 wave started as a disease in wealthy rural counties in southern Germany and ventured into poorer urban and agricultural counties during the course of the first wave. High age-standardised incidence in low socioeconomic status (SES) counties became more pronounced from the second lockdown period onwards, when wealthy counties appeared to be better protected. Features related to economic and educational characteristics of the young population in a county played an important role at the beginning of the pandemic up to the second lockdown phase, as did features related to the population living in nursing homes; those related to international migration and a large proportion of foreigners living in a county became important in the postlockdown period. CONCLUSION: High mobility of high SES groups may drive the pandemic at the beginning of waves, while mitigation measures and beliefs about the seriousness of the pandemic as well as the compliance with mitigation measures may put lower SES groups at higher risks later on.

Regional Characteristics of the Second Wave of SARS-CoV-2 Infections and COVID-19 Deaths in Germany.

(1) Background: In the absence of individual level information, the aim of this study was to identify the regional key features explaining SARS-CoV-2 infections and COVID-19 deaths during the upswing of the second wave in Germany. (2) Methods: We used COVID-19 diagnoses and deaths from 1 October to 15 December 2020, on the county-level, differentiating five two-week time periods. For each period, we calculated the age-standardized COVID-19 incidence and death rates on the county level. We trained gradient boosting models to predict the incidence and death rates by 155 indicators and identified the top 20 associations using Shap values. (3) Results: Counties with low socioeconomic status (SES) had higher infection and death rates, as had those with high international migration, a high proportion of foreigners, and a large nursing home population. The importance of these characteristics changed over time. During the period of intense exponential increase in infections, the proportion of the population that voted for the Alternative for Germany (AfD) party in the last federal election was among the top characteristics correlated with high incidence and death rates. (4) Machine learning approaches can reveal regional characteristics that are associated with high rates of infection and mortality.

[Retrospective results of radiation therapy of the eustachian tube in chronic otitis media]. / Retrospektive Ergebnisse der perkutanen Strahlentherapie der Tuba Eustachii bei chronischer Otitis media.

BACKGROUND: The treatment results of symptomatic radiation therapy of the Eustachian tube in chronic otitis media had to be evaluated retrospectively. PATIENTS AND METHODS: Between 1980 and 1997, 66 patients were referred for therapy. The median age was 58 years. In the clinical presentation, all the patients had a hearing impairment, 35 patients complained of pain, 21 had otorrhea. In their history, 20 patients indicated chronic recurrent infections. The complaints lasted for 4.7 years in the median, primary conservative (adstringentia, antibiotics) and surgical treatment (paracentesis, tympanic tubule, tympanoplastic) did not lead to lasting cure. In 40 of 66 patients, finally radiation therapy was done of both Eustachian tubes. With opposed fields and cobalt-60 photons a total dose of 6 Gy at single doses of 1 Gy, three times a week, was applied. Under the causes for exclusion of radiation therapy were non-acceptance of the patients (nine), prior radiation therapies (six) or spontaneous improvement after initial presentation in our department. The treatment results were evaluated by interviews of the patients and regular otorhinolaryngological examinations. RESULTS: There were no side effects noticed. 28 of 40 (70%) patients reported a significant improvement that could be verified by objective otorhinolaryngological examinations. In the group of 26 nonirradiated patients, 22 could be interviewed indicating in 16 cases (72%) that the complaints were unchanged and chronic otitis media was lasting. In a subgroup analysis concerning the duration of otitis media radiation therapy proved more effective in an acute and subacute stadium of disease of up to 5 years duration, while the patients resistant to radiation therapy were entirely in a chronic stage of disease exceeding 5 years duration. CONCLUSIONS: Radiation therapy is an effective tool for symptomatic improvement of the therapy-resistant chronic otitis media. A dose of 6 Gy seems to be sufficient to achieve an antiinflammatory effect. Radiotherapy should be applied earlier after initial conservative and surgical treatment.