Deriving Treatment Decision Support From Dutch Electronic Health Records by Exploring the Applicability of a Precision Cohort-Based Procedure for Patients With Type 2 Diabetes Mellitus: Precision Cohort Study.

BACKGROUND: The rapidly increasing availability of medical data in electronic health records (EHRs) may contribute to the concept of learning health systems, allowing for better personalized care. Type 2 diabetes mellitus was chosen as the use case in this study. OBJECTIVE: This study aims to explore the applicability of a recently developed patient similarity-based analytics approach based on EHRs as a candidate data analytical decision support tool. METHODS: A previously published precision cohort analytics workflow was adapted for the Dutch primary care setting using EHR data from the Nivel Primary Care Database. The workflow consisted of extracting patient data from the Nivel Primary Care Database to retrospectively generate decision points for treatment change, training a similarity model, generating a precision cohort of the most similar patients, and analyzing treatment options. This analysis showed the treatment options that led to a better outcome for the precision cohort in terms of clinical readouts for glycemic control. RESULTS: Data from 11,490 registered patients diagnosed with type 2 diabetes mellitus were extracted from the database. Treatment-specific filter cohorts of patient groups were generated, and the effect of past treatment choices in these cohorts was assessed separately for glycated hemoglobin and fasting glucose as clinical outcome variables. Precision cohorts were generated for several individual patients from the filter cohorts. Treatment options and outcome analyses were technically well feasible but in general had a lack of statistical power to demonstrate statistical significance for treatment options with better outcomes. CONCLUSIONS: The precision cohort analytics workflow was successfully adapted for the Dutch primary care setting, proving its potential for use as a learning health system component. Although the approach proved technically well feasible, data size limitations need to be overcome before application for clinical decision support becomes realistically possible.

Clinical impact of an integrated e-health system for diabetes self-management support and shared decision making (POWER2DM): a randomised controlled trial.

AIMS/HYPOTHESIS: There is a lack of e-health systems that integrate the complex variety of aspects relevant for diabetes self-management. We developed and field-tested an e-health system (POWER2DM) that integrates medical, psychological and behavioural aspects and connected wearables to support patients and healthcare professionals in shared decision making and diabetes self-management. METHODS: Participants with type 1 or type 2 diabetes (aged >18 years) from hospital outpatient diabetes clinics in the Netherlands and Spain were randomised using randomisation software to POWER2DM or usual care for 37 weeks. This RCT assessed the change in HbA1c between the POWER2DM and usual care groups at the end of the study (37 weeks) as a primary outcome measure. Participants and clinicians were not blinded to the intervention. Changes in quality of life (QoL) (WHO-5 Well-Being Index [WHO-5]), diabetes self-management (Diabetes Self-Management Questionnaire - Revised [DSMQ-R]), glycaemic profiles from continuous glucose monitoring devices, awareness of hypoglycaemia (Clarke hypoglycaemia unawareness instrument), incidence of hypoglycaemic episodes and technology acceptance were secondary outcome measures. Additionally, sub-analyses were performed for participants with type 1 and type 2 diabetes separately. RESULTS: A total of 226 participants participated in the trial (108 with type 1 diabetes; 118 with type 2 diabetes). In the POWER2DM group (n=111), HbA1c decreased from 60.6±14.7 mmol/mol (7.7±1.3%) to 56.7±12.1 mmol/mol (7.3±1.1%) (means ± SD, p<0.001), compared with no change in the usual care group (n=115) (baseline: 61.7±13.7 mmol/mol, 7.8±1.3%; end of study: 61.0±12.4 mmol/mol, 7.7±1.1%; p=0.19) (between-group difference 0.24%, p=0.008). In the sub-analyses in the POWER2DM group, HbA1c in participants with type 2 diabetes decreased from 62.3±17.3 mmol/mol (7.9±1.6%) to 54.3±11.1 mmol/mol (7.1±1.0%) (p<0.001) compared with no change in HbA1c in participants with type 1 diabetes (baseline: 58.8±11.2 mmol/mol [7.5±1.0%]; end of study: 59.2±12.7 mmol/mol [7.6±1.2%]; p=0.84). There was an increase in the time during which interstitial glucose levels were between 3.0 and 3.9 mmol/l in the POWER2DM group, but no increase in clinically relevant hypoglycaemia (interstitial glucose level below 3.0 mmol/l). QoL improved in participants with type 1 diabetes in the POWER2DM group compared with the usual care group (baseline: 15.7±3.8; end of study: 16.3±3.5; p=0.047 for between-group difference). Diabetes self-management improved in both participants with type 1 diabetes (from 7.3±1.2 to 7.7±1.2; p=0.002) and those with type 2 diabetes (from 6.5±1.3 to 6.7±1.3; p=0.003) within the POWER2DM group. The POWER2DM integrated e-health support was well accepted in daily life and no important adverse (or unexpected) effects or side effects were observed. CONCLUSIONS/INTERPRETATION: POWER2DM improves HbA1c levels compared with usual care in those with type 2 diabetes, improves QoL in those with type 1 diabetes, improves diabetes self-management in those with type 1 and type 2 diabetes, and is well accepted in daily life. TRIAL REGISTRATION: ClinicalTrials.gov NCT03588104. FUNDING: This study was funded by the European Union's Horizon 2020 Research and Innovation Programme (grant agreement number 689444).

A computational model of the effects of macronutrients absorption and physical exercise on hormonal regulation and metabolic homeostasis.

Regular physical exercise and appropriate nutrition affect metabolic and hormonal responses and may reduce the risk of developing chronic non-communicable diseases such as high blood pressure, ischemic stroke, coronary heart disease, some types of cancer, and type 2 diabetes mellitus. Computational models describing the metabolic and hormonal changes due to the synergistic action of exercise and meal intake are, to date, scarce and mostly focussed on glucose absorption, ignoring the contribution of the other macronutrients. We here describe a model of nutrient intake, stomach emptying, and absorption of macronutrients in the gastrointestinal tract during and after the ingestion of a mixed meal, including the contribution of proteins and fats. We integrated this effort to our previous work in which we modeled the effects of a bout of physical exercise on metabolic homeostasis. We validated the computational model with reliable data from the literature. The simulations are overall physiologically consistent and helpful in describing the metabolic changes due to everyday life stimuli such as multiple mixed meals and variable periods of physical exercise over prolonged periods of time. This computational model may be used to design virtual cohorts of subjects differing in sex, age, height, weight, and fitness status, for specialized in silico challenge studies aimed at designing exercise and nutrition schemes to support health.

Exploring dynamic metabolomics data with multiway data analysis: a simulation study.

BACKGROUND: Analysis of dynamic metabolomics data holds the promise to improve our understanding of underlying mechanisms in metabolism. For example, it may detect changes in metabolism due to the onset of a disease. Dynamic or time-resolved metabolomics data can be arranged as a three-way array with entries organized according to a subjects mode, a metabolites mode and a time mode. While such time-evolving multiway data sets are increasingly collected, revealing the underlying mechanisms and their dynamics from such data remains challenging. For such data, one of the complexities is the presence of a superposition of several sources of variation: induced variation (due to experimental conditions or inborn errors), individual variation, and measurement error. Multiway data analysis (also known as tensor factorizations) has been successfully used in data mining to find the underlying patterns in multiway data. To explore the performance of multiway data analysis methods in terms of revealing the underlying mechanisms in dynamic metabolomics data, simulated data with known ground truth can be studied. RESULTS: We focus on simulated data arising from different dynamic models of increasing complexity, i.e., a simple linear system, a yeast glycolysis model, and a human cholesterol model. We generate data with induced variation as well as individual variation. Systematic experiments are performed to demonstrate the advantages and limitations of multiway data analysis in analyzing such dynamic metabolomics data and their capacity to disentangle the different sources of variations. We choose to use simulations since we want to understand the capability of multiway data analysis methods which is facilitated by knowing the ground truth. CONCLUSION: Our numerical experiments demonstrate that despite the increasing complexity of the studied dynamic metabolic models, tensor factorization methods CANDECOMP/PARAFAC(CP) and Parallel Profiles with Linear Dependences (Paralind) can disentangle the sources of variations and thereby reveal the underlying mechanisms and their dynamics.

Real-time monitoring of drug laboratory test interactions: a proof of concept.

OBJECTIVES: For the correct interpretation of test results, it is important to be aware of drug-laboratory test interactions (DLTIs). If DLTIs are not taken into account by clinicians, erroneous interpretation of test results may lead to a delayed or incorrect diagnosis, unnecessary diagnostic testing or therapy with possible harm for patients. A DLTI alert accompanying a laboratory test result could be a solution. The aim of this study was to test a multicentre proof of concept of an electronic clinical decision support system (CDSS) for real-time monitoring of DLTIs. METHODS: CDSS was implemented in three Dutch hospitals. So-called 'clinical rules' were programmed to alert medical specialists for possible DLTIs based on laboratory test results outside the reference range in combination with prescribed drugs. A selection of interactions from the DLTI database of the Dutch society of clinical chemistry and laboratory medicine were integrated in 43 clinical rules, including 24 tests and 25 drugs. During the period of one month all generated DTLI alerts were registered in the laboratory information system. RESULTS: Approximately 65 DLTI alerts per day were detected in each hospital. Most DLTI alerts were generated in patients from the internal medicine and intensive care departments. The most frequently reported DLTI alerts were potassium-proton pump inhibitors (16%), potassium-beta blockers (11%) and creatine kinase-statins (11%). CONCLUSIONS: This study shows that it is possible to alert for potential DLTIs in real-time with a CDSS. The CDSS was successfully implemented in three hospitals. Further research must reveal its usefulness in clinical practice.

Sleep duration and lipid metabolism in patients with diabetes mellitus: from the POWER2DM study.

This study assesses the association between sleep duration and plasma lipid profiles in people with diabetes mellitus (DM). Sleep duration data were obtained in 91 patients from the POWER2DM study (NCT03588104). The patients were divided in tertiles, based on their sleep duration, and blood samples were obtained at the beginning and after 9 months. Significant differences were found, specifically, patients in Tertile 3 (≥ 7.51 h) showed lower plasma levels of high-density lipoprotein cholesterol HDL-c (p < 0.05), apolipoprotein A1 (apo-A1; p < 0.05) and low HDL-c/apo-A1 ratio (p < 0.05). This study shows that sleep duration is associated with plasma lipid profiles in people with DM.

A Predictive Model-Based Decision Support System for Diabetes Patient Empowerment.

The main objective of POWER2DM is to develop and validate a personalized self-management support system (SMSS) for T1 and T2 diabetes patients that combines and integrates i) a decision support system (DSS) based on leading European predictive personalized models for diabetes interlinked with predictive computer models, ii) automated e-coaching functionalities based on Behavioral Change Theories, and iii) real-time Personal Data processing and interpretation. The SMSS offers a guided workflow based on treatment goals and activities where a periodic review evaluates the patients progress and provides detailed feedback on how to improve towards a healthier, diabetes appropriate lifestyle.

Clinical usefulness of drug-laboratory test interaction alerts: a multicentre survey.

OBJECTIVES: Knowledge of possible drug-laboratory test interactions (DLTIs) is important for the interpretation of laboratory test results. Failure to recognize these interactions may lead to misinterpretation, a delayed or erroneous diagnosis, or unnecessary extra diagnostic tests or therapy, which may harm patients. The aim of this multicentre survey was to evaluate the clinical value of DLTI alerts. METHODS: A survey was designed with six predefined clinical cases selected from the clinical laboratory practice with a potential DLTI. Physicians from several departments, including internal medicine, cardiology, intensive care, surgery and geriatrics in six participating hospitals were recruited to fill in the survey. The survey addressed their knowledge of DLTIs, motivation to receive an alert and opinion on the potential influence on medical decision making. RESULTS: A total of 210 physicians completed the survey. Of these respondents 93% had a positive attitude towards receiving DLTI alerts; however, the reported value differed per case and per respondent's background. In each clinical case, medical decision making was influenced as a consequence of the reported DLTI message (ranging from 3 to 45% of respondents per case). CONCLUSIONS: In this multicentre survey, most physicians stated DLTI messages to be useful in laboratory test interpretation. Medical decision making was influenced by reporting DLTI alerts in each case. Alerts should be adjusted according to the needs and preferences of the receiving physicians.

Appetite ratings of foods are predictable with an in vitro advanced gastrointestinal model in combination with an in silico artificial neural network.

The expected increase of global obesity prevalence makes it necessary to have information about the effects of meal intakes on the feeling of appetite. Because human clinical studies are time and cost intensive, there is a need for a reliable alternative. The aim of this study was to develop and evaluate an in vitro-in silico technology to predict the feelings of fullness and hunger after consumption of different types of meals. In this technology the results from an in vitro gastrointestinal model (tiny-TIMagc) on gastric viscosity and intestinal digestion of different meals were used as input data for an in silico artificial neural network (ANN). The predictions of the feeling of fullness and hunger were compared with actual human scores for these parameters after intake of the same type of meals. From these first series of experiments, with a relatively small number of in vitro digestive parameters as input for in silico modeling, a reasonable prediction of appetite rating for foods can be realized at a time- and cost-effective way.

Diagnostic error as a result of drug-laboratory test interactions.

Background Knowledge of possible drug-laboratory test interactions (DLTIs) is important for the interpretation of laboratory test results. Test results may be affected by physiological or analytical drug effects. Failure to recognize these interactions may lead to misinterpretation of test results, a delayed or erroneous diagnosis or unnecessary extra tests or therapy, which may harm patients. Content Thousands of interactions have been reported in the literature, but are often fragmentarily described and some papers even reported contradictory findings. How can healthcare professionals become aware of all these possible interactions in their individual patients? DLTI decision support applications could be a good solution. In a literature search, only four relevant studies have been found on DLTI decision support applications in clinical practice. These studies show a potential benefit of automated DLTI messages to physicians for the interpretation of laboratory test results. All physicians reported that part of the DLTI messages were useful. In one study, 74% of physicians even sometimes refrained from further additional examination. Summary and outlook Unrecognized DLTIs potentially cause diagnostic errors in a large number of patients. Therefore, efforts to avoid these errors, for example with a DLTI decision support application, could tremendously improve patient outcome.

Determination of physiological parameters for endogenous glucose production in individuals using diurnal data.

BACKGROUND: Triple tracer meal experiments used to investigate organ glucose-insulin dynamics, such as endogenous glucose production (EGP) of the liver are labor intensive and expensive. A procedure was developed to obtain individual liver related parameters to describe EGP dynamics without the need for tracers. RESULTS: The development used an existing formula describing the EGP dynamics comprising 4 parameters defined from glucose, insulin and C-peptide dynamics arising from triple meal studies. The method employs a set of partial differential equations in order to estimate the parameters for EGP dynamics. Tracer-derived and simulated data sets were used to develop and test the procedure. The predicted EGP dynamics showed an overall mean R 2 of 0.91. CONCLUSIONS: In summary, a method was developed for predicting the hepatic EGP dynamics for healthy, pre-diabetic, and type 2 diabetic individuals without applying tracer experiments.

Impact of interactions between drugs and laboratory test results on diagnostic test interpretation - a systematic review.

Intake of drugs may influence the interpretation of laboratory test results. Knowledge and correct interpretation of possible drug-laboratory test interactions (DLTIs) is important for physicians, pharmacists and laboratory specialists. Laboratory results may be affected by analytical or physiological effects of medication. Failure to take into account the possible unintended influence of drug use on a laboratory test result may lead to incorrect diagnosis, incorrect treatment and unnecessary follow-up. The aim of this review is to give an overview of the literature investigating the clinical impact and use of DLTI decision support systems on laboratory test interpretation. Particular interactions were reported in a large number of articles, but they were fragmentarily described and some papers even reported contradictory findings. To provide an overview of information that clinicians and laboratory staff need to interpret test results, DLTI databases have been made by several groups. In a literature search, only four relevant studies have been found on DLTI decision support applications for laboratory test interpretation in clinical practice. These studies show a potential benefit of automated DLTI messages to physicians for the correct interpretation of laboratory test results. Physicians reported 30-100% usefulness of DLTI messages. In one study 74% of physicians sometimes even refrained from further additional examination. The benefit of decision support increases when a refined set of clinical rules is determined in cooperation with health care professionals. The prevalence of DLTIs is high in a broad range of combinations of laboratory tests and drugs and these frequently remain unrecognized.

Actions of metformin and statins on lipid and glucose metabolism and possible benefit of combination therapy.

Patients with diabetes type 2 have an increased risk for cardiovascular disease and commonly use combination therapy consisting of the anti-diabetic drug metformin and a cholesterol-lowering statin. However, both drugs act on glucose and lipid metabolism which could lead to adverse effects when used in combination as compared to monotherapy. In this review, the proposed molecular mechanisms of action of statin and metformin therapy in patients with diabetes and dyslipidemia are critically assessed, and a hypothesis for mechanisms underlying interactions between these drugs in combination therapy is developed.

Correlations between plasma strontium concentration, components of calcium and phosphate metabolism and renal function in type 2 diabetes mellitus.

BACKGROUND: Renal function decline in diabetic kidney disease is accompanied by calcium and phosphate metabolism alterations. Whereas strontium (Sr2+ ) has many similarities with calcium, little is known about Sr2+ in this respect. We studied the association of plasma Sr2+ concentration and parameters associated with an altered calcium and phosphate metabolism in diabetic kidney disease. MATERIALS AND METHODS: Plasma Sr2+ concentration was measured in 450 patients included in the DIAbetes and LifEstyle Cohort Twente-1. Patients were classified based on chronic kidney disease (CKD) stages: stages 1-2, stage 3 and stages 4-5 (estimated glomerular filtration rate of ≥60 mL·min-1 ·1.73 m-2 , 30-59 mL·min-1 ·1.73 m-2 and ≤29 mL·min-1 ·1.73 m-2 , respectively). The associations between log-transformed plasma Sr2+ concentration and parameters of calcium and phosphate metabolism were studied using multivariate linear regression analysis. RESULTS: Overall, median plasma Sr2+ concentration was in normal range, 269 nmol/L, but was progressively higher in patients with lower renal function, that is 246 nmol/L (CKD 1-2), 347 nmol/L (CKD 3) and 419 nmol/L (CKD 4-5). In multivariate analysis, independent associations were found between plasma Sr2+ concentration and both eGFR (ß = -0.401, P < 0.001) and plasma fibroblast growth factor 23 (FGF23) concentration (ß = 0.087, P = 0.04). CONCLUSIONS: We found an independent inverse association between eGFR and plasma Sr2+ concentration and an independent association between plasma Sr2+ concentration and plasma FGF23 concentration, a marker of deranged calcium and phosphate metabolism. Further research is needed to determine the mechanisms behind these associations and the impact of an elevation in plasma Sr2+ concentration on bone mineralization and calcification.

Feed-forward neural network model for hunger and satiety related VAS score prediction.

BACKGROUND: An artificial neural network approach was chosen to model the outcome of the complex signaling pathways in the gastro-intestinal tract and other peripheral organs that eventually produce the satiety feeling in the brain upon feeding. METHODS: A multilayer feed-forward neural network was trained with sets of experimental data relating concentration-time courses of plasma satiety hormones to Visual Analog Scales (VAS) scores. The network successfully predicted VAS responses from sets of satiety hormone data obtained in experiments using different food compositions. RESULTS: The correlation coefficients for the predicted VAS responses for test sets having i) a full set of three satiety hormones, ii) a set of only two satiety hormones, and iii) a set of only one satiety hormone were 0.96, 0.96, and 0.89, respectively. The predicted VAS responses discriminated the satiety effects of high satiating food types from less satiating food types both in orally fed and ileal infused forms. CONCLUSIONS: From this application of artificial neural networks, one may conclude that neural network models are very suitable to describe situations where behavior is complex and incompletely understood. However, training data sets that fit the experimental conditions need to be available.

The effect of plant sterols and different low doses of omega-3 fatty acids from fish oil on lipoprotein subclasses.

SCOPE: Consumption of a low-fat spread enriched with plant sterols (PS) and different low doses (<2 g/day) of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) from fish oil reduces serum triglycerides (TGs) and low-density lipoprotein-cholesterol (LDL-Chol) and thus beneficially affects two blood lipid risk factors. Yet, their combined effects on TG and Chol in various lipoprotein subclasses have been investigated to a limited extent. METHODS AND RESULTS: In a randomized, double-blind, placebo-controlled, parallel study, we determined TG and Chol in 13 LP subclasses in fasting serum of 282 hypercholesterolemic subjects, who consumed either a placebo spread or one of the four spreads containing PS (2.5 g/day) and EPA+DHA (0.0, 0.9, 1.3, and 1.8 g/day) for 4 weeks. After PS treatment, total LDL-Chol was reduced, which was not further changed by EPA+DHA. No shift in the LDL-Chol particle distribution was observed. The addition of EPA+DHA to PS dose-dependently reduced VLDL-Chol and VLDL-TG mainly in larger particles. Furthermore, the two highest doses of EPA+DHA increased Chol and TG in the larger HDL particles, while these concentrations were decreased in the smallest HDL particles. CONCLUSION: The consumption of a low-fat spread enriched with both PS and EPA+DHA induced shifts in the lipoprotein distribution that may provide additional cardiovascular benefits over PS consumption alone.