Category: Family

Personalized diabetes care

Personalized diabetes care

Tailoring Perzonalized modules fiabetes content to different professional and cultural settings is ideally suited to these partner organizations. Probiotics for gut health is important to note that these set points are arbitrary and Effective fat loss an diaetes of the complex decision-making process Personwlized is often Personalizfd in diabetes care. Heterozygous Personalized diabetes care mutations Persohalized present in approximately half Personalizrd Probiotics for gut health ccare develop diabetes within the first 6 months after birth, making this a subgroup of diabetes patients who can readily be selected for genetic testing. Identification of HKDC1 and BACE2 as genes influencing glycemic traits during pregnancy through genome-wide association studies. Warnings - Do not take Lyumjev or Humalog if you have: symptoms of low blood sugar hypoglycemia an allergy to insulin lispro-aabc, Humalog, or any of the ingredients in Lyumjev or Humalog. Using this technology, physicians can view patient data in real time and between clinic visits and incorporate their lifelog data with clinical data derived from personal sensors and wearables in electronic medical records EMRs. d SBP: systolic blood pressure.

Personalized diabetes care -

In addition, lifestyle changes based on PA and diet records; cardiometabolic risk factors such as body weight, blood pressure, and lipid profile; program satisfaction and compliance or adherence ; frequency of hypoglycemia; and changes in homeostasis model assessment of insulin resistance and β cell function were assessed at 26 weeks.

Adherence was defined as the proportion of intervention participation using the iCareD app, including blood glucose measurement and feedback confirmation, over a week period. Exploratory assessment variables included changes in diabetes prescriptions, SMBG frequency, and BMI.

Participant satisfaction was assessed in the 2 intervention groups by using a locally developed satisfaction survey at 26 weeks. The survey included 5-level Likert-type questions evaluating self-care efficacy and various opinions on the iCareD system, such as the ease and frequency of text messages, perceived efficacy, and willingness to continue with or recommend the iCareD program to family or friends.

A score of 5 indicated very satisfied or strongly agree. Higher scores on the satisfaction scale reflect better results.

Demographic and clinical information collected at baseline and follow-up has been described previously [ 13 ]. PA was tracked using a Google Fit mobile app and assessed as the total step count per day [ 17 ]. Body composition data were obtained using a bioimpedance analyzer InBody and , InBody Co, Ltd at baseline and every 26 weeks.

Laboratory parameters, including fasting glucose, HbA 1c level, and lipid profile, were collected at every visit. C-peptide and urinary albumin to creatinine ratios were measured at baseline and every 26 weeks.

We used the updated homeostasis model assessment calculator to evaluate the homeostasis model assessment of insulin resistance and β cell function [ 18 - 20 ]. Diabetes management behaviors such as SMBG frequency, PA, and diet records were obtained at every visit.

SMBG frequency was defined as the average number of tests performed per day, calculated for each patient based on the records in the web system.

User satisfaction with mobile app was surveyed in the MC and MPC groups. Continuous variables were presented as mean SD , whereas categorical data were presented as frequencies with percentages.

Analysis of covariance was used to compare the mean week HbA 1c levels among the 3 groups. Post hoc analysis was performed using the Bonferroni method. The number of hypoglycemic events among the groups was compared using the chi-square test or Fisher exact test.

Missing data were replaced by the last-observation-carried-forward method for all participants who were followed up at least once after enrollment. Both per-protocol and ITT analyses were conducted. Unless otherwise specified, analyses were performed based on the results of the ITT analysis.

The analysis was performed using SAS version 9. The study protocol was approved by the ethics committee of St. All participants provided written informed consent before enrollment in the study. All data and information were anonymized according to the International Conference on Harmonization Good Clinical Practice guidelines.

During the recruitment period from August to August in the outpatient clinics of 2 separate university-affiliated diabetes centers, a total of participants were assessed for eligibility and A total of 10 participants withdrew consent, leaving participants to be included in this study Figure 2.

After the week follow-up, the total retention rate was The baseline analysis revealed no significant differences between those who completed the study and those who were lost to follow-up data not shown. The mean age of the participants was The mean baseline HbA 1c level and duration of diabetes were 8.

The mean BMI was None of the other baseline characteristics or variables differed significantly among the 3 study groups.

c MPC: mobile diabetes self-care with personalized, bidirectional feedback from physicians. The change in HbA 1c levels did not differ significantly at 26 weeks among the 3 groups Figure 3.

In the post hoc analysis, only the MPC group showed a significant decrease in HbA 1c levels compared with the UC group.

Table 2. Adjusting for age, sex, and baseline HbA 1c level did not affect the HbA 1c level change results. Other changes in clinical and behavioral outcomes from baseline to follow-up are shown in Table 3.

However, the change in fasting glucose levels did not differ among the 3 groups during the week intervention period Table 3. Changes in body weight and BMI from baseline to 26 weeks also showed no differences among the study groups.

The frequency of the SMBG did not show any significant differences among the 3 groups at the week follow-up. However, compared with patients in the UC group, those in the 2 intervention groups iCareD system users tended to have more frequent SMBG recordings at 12 weeks UC vs iCareD system users: 1.

PA, defined as step counts per day, was not significantly different among the study groups 26 weeks after the intervention. The goal achievement rate for PA was higher in the MC and MPC groups than that in the UC group at 26 weeks, but the difference was not significant UC vs MC vs MPC: Low-density lipoprotein—cholesterol levels increased in the UC group and decreased in the MC and MPC groups during the follow-up period.

A total of out of the No differences were observed between the 2 groups Table 4. b MPC: mobile diabetes self-care with personalized, bidirectional feedback from physicians. There were no statistically significant differences between the MC and MPC groups in skill and technique acquisition, health service navigation, or manipulation of app content.

An end-of-intervention usability survey demonstrated that participants were comfortable with using the iCareD system. With regard to adherence, compared with participants in the MC group, those in the MPC group checked the automated text messages from the iCareD system for 26 weeks.

No serious adverse events were reported from enrollment until the completion of this study. Hypoglycemic events were infrequent and showed no differences among the groups at 26 weeks Table 5. No deaths, direct study-related adverse events, or severe hypoglycemic episodes were reported or detected.

This study was an RCT investigating a hospital-based, EMR-integrated mobile app—based diabetes self-care intervention over a week period in patients with T2DM. Although the HbA 1c level decreased from the baseline value in all 3 groups, the HbA 1c changes did not show any significant difference between the control and 2 intervention groups at 26 weeks.

Owing to the global growth in the use of mobile phones with powerful platforms to help health care, many types of apps have been developed. In , more than diabetes-related apps were reported to be available to users.

Mobile apps related to diabetes management generally deal with information about diabetes, healthy diet, PA, weight loss, the SMBG, adherence, and motivation [ 6 ]. mHealth interventions support self-care and diabetes education and encourage lifestyle modification.

These data may be used to tailor feedback messages or advice on specific behavior changes to implement; these messages are usually sent automatically according to an algorithm [ 5 , 9 , 23 ].

Compared with conventional mobile apps that collect only patient-driven data, our EMR-integrated mobile app could provide important clues to the future direction of mobile app development for diabetes management in 2 respects.

Given the high rates of comorbidity and concurrent medications in patients with T2DM [ 24 ], this integrated provision of medical information may allow HCPs to provide accurate guidance to patients on diet, exercise, and management of comorbid diseases rather than simply focusing on the message to lower blood glucose levels.

In particular, our systems adopted visualization of glucose levels by color to improve awareness or alertness of hyperglycemia red or hypoglycemia black [ 13 ]. Using the EMR-integrated mobile app intervention, we demonstrated a significant reduction in HbA 1c levels after 12 weeks of intervention.

Consistent with our results, a 3-month RCT using DialBetics, a smartphone-based self-management support system for Japanese patients with T2DM, demonstrated that HbA 1c levels decreased by an average of 0. A systematic review also revealed limited robust evidence of the promising short-term effectiveness of mHealth interventions for diabetes, such as the improvement of HbA 1c levels [ 14 , 15 , 26 , 27 ].

However, a caveat of these RCT analyses is that most of them included only studies conducted under highly controlled conditions with a small number of patients [ 14 , 15 , 26 , 27 ]. It is noteworthy that our study showed significant differences in HbA 1c levels among groups at 3 months in real-world practice, with a relatively large number of patients at 2 different clinical sites.

This finding suggests the potential usefulness of EMR-integrated mobile app interventions in diabetes management. In addition, we found that the intervention effects in the MPC group were prominent in patients with younger age, obesity, higher C-peptide levels, and no insulin treatment.

This finding implies that mobile-based interventions, such as other diabetes treatments, may be more effective when β cell function is preserved. This also highlights the importance of early intervention.

This indicates that although early intervention may be important, such interventions may also be effective in long-standing diabetes. Mobile phone apps that receive blood glucose data from a connected glucometer are available and have the capacity to make data upload and review less burdensome [ 30 ].

The internet-based SMBG system, which augments the SMBG by giving patients the means to communicate their blood glucose levels to their HCP for actional feedback, has been shown to reduce HbA 1c levels in some RCTs involving patients with T2DM [ 31 ].

The inverse correlation between reporting frequency and HbA 1c levels, as well as the significant difference in HbA 1c levels only for frequent testers defined as those who test on average twice or more per day , suggests that frequent SMBG has an effect on reducing HbA 1c levels only when combined with regular, frequent communication of SMBG with an HCP [ 32 ].

The recording of a food diary using a smartphone app is a well-known simple tool, and technology to use images to quantify the composition and calorie content of food has been developed. However, it is difficult and cumbersome for users to constantly record data based on their eating habits [ 33 , 34 ].

However, automated integration of glucose and lifelog data in the EMR between scheduled clinic visits improves the HCP workflow for reviewing data and improves communication with patients, eventually leading to better care [ 35 ].

In this study, There are possible explanations for the lack of improvement in HbA 1c levels in mHealth app users. First, age is a barrier to digital health care adoption and may influence the adoption of new technologies [ 36 ]. The mean age of the participants in this study was Second, the iCareD system was developed with a focus on lifestyle changes rather than strict glucose control or active medication adjustment, such as whirlwind dosage escalation of antidiabetic medications.

In the case of the TExT-MED study, a unidirectional text message intervention for diabetes self-care providing text message triggers to encourage individuals to engage in self-care behaviors, the TExT-MED program also did not result in a significant improvement in HbA 1c levels.

However, trends toward improvement in the primary outcome of HbA 1c levels and other secondary outcomes, including quality of life, were observed. Patient engagement was highest for more medical topics, such as glucose monitoring and medications, and lower for lifestyle topics, such as PA and healthy coping [ 6 ].

Therefore, we suggest that interventions for diabetes self-care should include improving HbA 1c levels through modification of lifestyle, glucose monitoring, and adherence and dosage adjustment for antidiabetic medications [ 37 ].

Third, our patients had a long duration of diabetes and were insulin users [ 32 ]. In general, the effects of education and lifestyle changes decrease with the duration of diabetes [ 38 ].

Fourth, there was no evidence of the most effective frequency of the intervention messages. We sent personalized intervention messages from HCPs every 2 weeks and automated general informative messages every other day.

Patient satisfaction and accessibility are important for improving self-management efficiency, and the clinical course can be improved through personalized intervention [ 4 ]. More frequent, bidirectional, real-time communication with HCPs and patients would lead to more effective improvement in HbA 1c levels.

Although we did not observe remarkable improvement in HbA 1c levels over the long term, it is encouraging that the goal achievement rates for PA were higher in the intervention group at 26 weeks.

When the target of steps per day was applied [ 39 - 41 ], the difference in goal achievement rates among the groups further increased UC vs MC vs MPC: Given the lifelong management of T2DM, the small differences observed in the short term may increase in the future.

Furthermore, in terms of the prevention of diabetic complications such as cardiovascular disease, PA cannot be overemphasized [ 40 - 42 ]. Finally, we expect that our study will provide more solid evidence of the short-term efficacy of mobile app—based diabetes management.

In particular, in relation to the recent global public health crisis, the COVID pandemic, this methodology is expected to contribute greatly in the future to promote the rapid introduction and diffusion of new digital health—related technologies such as telemedicine [ 43 ].

To maximize the effect of mHealth interventions, it is important to tailor the intervention in a patient-centered manner and evaluate user satisfaction [ 44 ]. Undoubtedly, more RCTs with longer follow-up periods should be conducted to evaluate the long-term effects of diabetes-related mobile apps and to confirm that the outcomes seen in initial studies are sustainable over time [ 22 ].

In summary, the use of iCareD apps for diabetes self-care can be considered an effective measure, especially when patients can communicate with HCPs [ 8 ]. Remote health data monitoring and real-time communication with patients supported self-care of diabetes, resulting in short-term improvement in HbA 1c levels.

An mHealth system for patients with T2DM should be developed to support and motivate sustainable behavior changes in patients and to allow for an approach that is more tailored to individual needs.

The funders did not play any role in the study design, data collection and analysis, management preparation, or decision to publish.

EYL contributed to the drafting and revision of the manuscript, supervision of the study, and acquisition of data. JSY and SAC contributed to the revision of the manuscript, supervision of the study, and acquisition of data. SYL and JHL contributed to the statistical methodology and data management.

YBA and KHY contributed to the revision of the manuscript and supervision of the study. SHK contributed to the revision of the manuscript and design and supervision of the study.

All authors have read and approved the final manuscript. Hemoglobin A 1c HbA 1c level and HbA1c level changes from baseline to 12 and 26 weeks according to subgroup analysis. Edited by R Kukafka; submitted org , Skip to Main Content Skip to Footer. Efficacy of Personalized Diabetes Self-care Using an Electronic Medical Record—Integrated Mobile App in Patients With Type 2 Diabetes: 6-Month Randomized Controlled Trial Efficacy of Personalized Diabetes Self-care Using an Electronic Medical Record—Integrated Mobile App in Patients With Type 2 Diabetes: 6-Month Randomized Controlled Trial Authors of this article: Eun Young Lee 1 ; Seon-Ah Cha 2 ; Jae-Seung Yun 3 ; Sun-Young Lim 4 ; Jin-Hee Lee 4 ; Yu-Bae Ahn 3 ; Kun-Ho Yoon 1 ; Min Kyung Hyun 5 ; Seung-Hyun Ko 3.

Article Authors Cited by 8 Tweetations 4 Metrics. Original Paper. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea 2 Division of Endocrinology and Metabolism, Department of Internal Medicine, Wonkwang University Sanbon Hospital, Gunpo, Republic of Korea 3 Division of Endocrinology and Metabolism, Department of Internal Medicine, St.

Corresponding Author: Seung-Hyun Ko, MD, PhD Division of Endocrinology and Metabolism, Department of Internal Medicine, St. type 2 diabetes mellitus ; digital health ; mobile health ; mHealth ; mobile app ; self-monitoring blood glucose ; mobile phone.

Introduction Background Diabetes is one of the most important chronic diseases that threatens public health [ 1 ]. Objectives Therefore, we designed a diabetes management system using an EMR-integrated mobile app that provided regular feedback from HCPs to support diabetes self-care in a clinical setting for patients with T2DM.

Methods Study Design This study was designed as a week open-label, parallel group, 3-arm RCT conducted in 2 separate university-affiliated hospitals from August to December Study design.

HbA 1c : hemoglobin A 1c. Flowchart of patient enrollment and status. MC: mobile diabetes self-care; MPC: mobile diabetes self-care with personalized, bidirectional feedback from physicians; UC: usual care. Sample Size On the basis of previous studies [ 10 , 14 , 15 ], we assumed a mean difference in HbA 1c levels of at least 0.

Randomization Those who signed the informed consent form were randomly assigned to 1 of 3 groups in a ratio. Intervention Regardless of the assigned group, all participants were provided a glucometer CareSens N; i-SENS, Inc from which SMBG data were automatically transferred to our mHealth system, called the iCareD system.

Measurements Demographic and clinical information collected at baseline and follow-up has been described previously [ 13 ]. Statistical Analysis Continuous variables were presented as mean SD , whereas categorical data were presented as frequencies with percentages.

Ethics Approval The study protocol was approved by the ethics committee of St. Results Participant Flow During the recruitment period from August to August in the outpatient clinics of 2 separate university-affiliated diabetes centers, a total of participants were assessed for eligibility and Clinical Characteristics of Participants The mean age of the participants was Table 1.

Baseline demographic and clinical characteristics of patients. b MC: mobile diabetes self-care. d SBP: systolic blood pressure. e DBP: diastolic blood pressure.

f CVD: cardiovascular disease. g eGFR: estimated glomerular filtration rate. h HbA 1c : hemoglobin A 1c. i HDL: high-density lipoprotein. j LDL: low-density lipoprotein. k HOMA-IR: homeostasis model assessment for insulin resistance.

l HOMA-β: homeostasis model assessment for β-cell function. Table 3. Secondary study outcomes at baseline and follow-up. d PA: physical activity. e LDL: low-density lipoprotein. f HOMA-IR: homeostasis model assessment for insulin resistance.

g HOMA-β: homeostasis model assessment for β-cell function. Table 4. Messages read for 6 months intervention period and program satisfaction. If the patient received oral blood glucose regulation agents subsequent to one of these diagnosis codes, we assumed the diagnosis record was an error.

Neither the size of the population nor the proportion with good glycemic control changed substantially over the course of the study. Individual patients with longer medical histories may be overrepresented in the sample. We divided each patient's medical history into distinct lines of therapy, each characterized by a particular drug regimen Supplementary Fig.

Within each line of therapy, we considered patient visits occurring every days, corresponding to the life cycle of a red blood cell These patient visits provided the basis for our definition of patient outcomes.

We developed an algorithm to define precisely when each line of therapy ends and the next line begins according to when the combination of drugs prescribed to the patient changes in the EMR data. Each line of therapy was characterized by a unique drug regimen, defined to include all blood glucose regulation agents prescribed to the patient within the first 6 months after starting that line of therapy.

Regimens were defined as combinations of drugs from one or more drug classes. The drug classes we considered were metformin, insulin, and other blood glucose regulation agents; the other agents included sulfonylureas, thiazolidinediones, dipeptidyl peptidase 4 inhibitors, meglitinides, α-glucosidase inhibitors, glucagon-like peptide 1 agonists, and other antihyperglycemic agents.

A combination of drug classes was included as a regimen type if it was observed in a sufficient number of patient visits. This definition of end date for each line of therapy intends to capture the period when the patient was experiencing the effect of the drug regimen. Within each line of therapy, we considered patient visits occurring every days, beginning with the visit at which that regimen was initiated and continuing until no later than 80 days prior to the start of the subsequent regimen.

There were 48, unique patient visits in our dataset Table 1. At each visit, we defined a set of visit-specific patient characteristics, including the current line of therapy i. The outcome was measured as average HbA 1c 75 to days after the visit.

This effect period was chosen to allow for a complete red blood cell life cycle to elapse before measuring the effect of a drug therapy. We defined the standard of care for each visit as the drug regimen that was administered.

For The menu of options for a given patient could be determined by the provider, accounting for contraindications and other preferences, such as not using intensive control for elderly patients or patients with a history of severe hypoglycemia.

Specifically, the algorithm considered only regimens that represented an incremental addition or subtraction of a drug, or substitution of a drug of comparable intensity; metformin and insulin were considered to be of the lowest and highest intensity, respectively.

The menu options used in our analysis, differentiated by current treatment, are depicted in Fig. HbA 1c benefit of prescriptive algorithm for patients switching regimens. Each cell in the figure represents patients for whom the prescriptive algorithm recommended switching from the regimen on the vertical axis to the regimen on the horizontal axis.

Each cell is labeled with the number of patients who made that switch; cells labeled with a dash were not on the menu of options provided to patients currently on a given regimen.

For each patient visit, the outcomes predicted by k NN under each treatment were compared. We chose the optimal threshold value of 0. Increasing the threshold causes the algorithm to recommend switching for fewer patients, but the mean benefit among those who switch increases.

Above a certain threshold, the recommendation fits to noise in the training data and does not provide better mean benefits in the testing set. The optimal threshold balances these concerns. k NN regression is a nonparametric, instance-based algorithm that makes predictions by averaging the outcomes for the subset of observations most similar to the target as defined by some distance metric To predict potential outcomes under each regimen, we used a k NN regression based on a treatment-specific weighted Euclidean distance across normalized patient and visit-specific factors.

The weights were derived by training a separate ordinary least squares linear regression model for each treatment regimen and using the magnitudes of the regression coefficients Supplementary Table 1.

This weighted distance improves upon classical k NN by selecting neighbors based on the factors most predictive of HbA 1c outcome, rather than weighting all factors equally.

We considered factors from the following categories: demographic information, medical history, and treatment history. Specifically, the demographic factors used in the model were age, sex, and race. The prediction step of our algorithm is best illustrated through an example.

To identify the importance of each factor in predicting outcomes, we used patient visits in which metformin monotherapy was prescribed to train an ordinary least squares regression on normalized values of each patient factor listed above.

Thus, for any choice of k , we could rank the k closest neighbors from this treatment group. Intuitively, the number of neighbors k used to estimate posttreatment HbA 1c levels should increase with the size of the dataset.

For each treatment t , we found the value that minimized the root mean square error of the k NN predictions on a subset of the data not used to evaluate the algorithm.

To verify the accuracy of the k NN HbA 1c predictions, we evaluated the R 2 metric. Positive values of R 2 suggest patient characteristics are predictive of future HbA 1c.

For comparison, we evaluated the predictive accuracy of least absolute shrinkage and selection operator LASSO regression 14 and random forest 15 , two state-of-the-art machine-learning methods used widely because of their high prediction accuracy.

We used the predictions from these models in two alternative prescriptive algorithms. Because counterfactual treatment effects are not observable, we used the weighted matching approach embedded in the k NN regression to impute potential outcomes, an approach commonly used for causal inference in observational studies when randomization is unavailable For each visit, we applied our prescriptive algorithm to recommend a therapy.

If that recommendation matched the prescribed standard of care therapy, we observed the true effect from the therapy. Otherwise, the outcome was imputed by averaging the outcomes of the most similar patient visits at which the recommended therapy was administered; these similar visits were chosen from a test set not used for training, and the number of neighbors was selected to fit the size of the test set.

This estimated outcome was compared with the true outcome under standard of care at the given patient visit. Our hypothesis was that the average predicted HbA 1c outcome after applying our prescriptive algorithm would be less than that observed from administering standard of care, resulting in a net average improvement in outcomes.

All analyses were performed in R 3. The R 2 of the k NN predictions on unseen data ranged from 0. The R 2 values from the LASSO and random forest models ranged from 0. The predictive power was similar across the three methods.

The performance of the prescriptive algorithm is summarized in Table 2. The mean HbA 1c outcome after treatment was 0. Of the 48, patient visits in our dataset, the algorithm differed from the standard of care for 15, visits, For this subset of visits, the mean HbA 1c outcome under the algorithm was lower by 0.

The median outcome for these visits was 0. For comparison, the median difference for all visits was zero because, for In our analysis, the mean difference in HbA 1c was more negative than the median because of a left-skewed distribution.

Some patients received particularly large benefits from using the prescriptive algorithm, which had an outsize effect on the mean but did not affect the median. Among trajectories with at least patients, the largest benefit of the algorithm was achieved through personalized recommendations for 7, patients currently on insulin monotherapy to switch to monotherapy with metformin or another blood glucose regulation agent.

However, for the vast majority of patients currently on insulin-based regimens, the algorithm recommends that those patients continue with that therapy. Among the 7, patient visits, those who were recommended to switch from insulin were on average younger mean The performance of the prescriptive algorithm in specific patient subgroups is summarized in Supplementary Table 3.

The overall mean HbA 1c outcome using the prescriptive algorithm was 0. The benefit of using the algorithm was 0. The benefit of the algorithm was 0. The benefit was 0. Our methodology motivates a provider dashboard that would report information on the demographics, medical history, and response to treatment for patients similar to an index patient.

A prototype dashboard visualization for one sample patient visit is shown in Fig. In addition, the dashboard would display the mean, SD, and full distribution of HbA 1c outcomes among the nearest neighbors who received each treatment in the menu of options. Based on this evidence, the dashboard would display a treatment recommendation.

The provider would have the ability to override this recommendation given any special management needs of the patient. For instance, if the patient is elderly and the distribution of HbA 1c outcomes indicates that the recommended therapy has an elevated risk of hypoglycemia, the provider may opt for an alternative treatment.

Visualization of prescriptive algorithm: provider dashboard prototype. This figure visualizes how the prescriptive algorithm can be used by providers for a single patient. In each subpanel in panel B , the posttreatment HbA 1c level is on the horizontal axis, and the number of visits is on the vertical axis.

D : Depicts the history of diabetes progression and treatment for the patient, with date along the horizontal axis. The vertical axis of the top subpanel indicates various drug classes; the vertical axis of the bottom subpanel depicts HbA 1c percentage.

The overall mean HbA 1c outcome using the LASSO-based prescriptive algorithm was lower by 0. The benefit from using the random forest—based prescriptive algorithm relative to standard of care was 0.

In the sensitivity analyses, under three alternate random splittings of the dataset, the overall mean benefit of using the prescriptive algorithm compared with standard of care ranged from 0.

To our knowledge, we present the first prescriptive method for personalized type 2 diabetes care. Using historical data from a large EMR database, this novel prescriptive method resulted in an average HbA 1c benefit of 0.

Our method incorporates patient-specific demographic and medical history data to determine the best course of treatment. Compared with other machine-learning methods considered, the k NN prescriptive approach is highly interpretable and flexible in clinical applications. The novelty of our approach is in personalizing the decision-making process by incorporating patient-specific factors.

This method can easily accommodate alternative disease-management approaches within specific subpopulations, such as patients with chronic kidney disease and elderly patients.

We believe this personalization is the primary driver of benefit relative to standard of care. In practice, the algorithm can be integrated into existing EMR systems to dynamically suggest personalized treatment paths for each patient based on historical records.

The algorithm ingests and analyzes EMR data and generates recommendations. An intuitive, interactive dashboard summarizes the evidence for the recommendation, including the expected distribution of outcomes under alternative treatments Fig.

Because of the nature of retrospective data from existing EMR, this study has several limitations. Patients were not randomized into treatment groups. Although our matching methodology controls for several confounding factors that could explain differences in treatment effects, we can only estimate counterfactual outcomes.

EMR data do not include socioeconomic factors or patient preferences that may be important in treatment decisions. Becaue of a lack of sufficient data, glucagon-like peptide 1 agonists were not considered as a separate drug class.

If more data were available, we could further differentiate regimen types beyond the 13 we include in this analysis. In addition, the study population from BMC may not be representative of the U.

population as a whole. With EMR medication-order data alone, we cannot be certain whether a prescribed medication was filled or taken and cannot know precisely when the medication was stopped.

Although this data quality issue could hamper attempts to make drug efficacy comparisons, our analysis aims to address the question of which drugs to prescribe under real-world scenarios. We optimize for an outcome that takes into account unobserved factors such as nonadherence.

For instance, if nonadherence is more prevalent among patients prescribed insulin than other regimens, this perspective may explain why, in our study population, the algorithm recommends insulin less often than it is prescribed in clinical practice.

Our method can be extended to be more flexible and comprehensive. Currently, the prescriptive algorithm does not support individualized glycemic targets; we assume that a lower glycemic level is always preferred. The study currently optimizes only for a single health outcome; a more comprehensive algorithm would consider adverse event outcomes as well.

Despite these limitations, the study establishes strong evidence of the benefit of individualizing diabetes care. The success of this data-driven approach invites further testing using datasets from other hospital and care settings. Testing the prescriptive algorithm in a clinical trial setting would provide even stronger evidence of clinical effectiveness.

As large-scale genomic data become more widely available, the algorithm could readily incorporate such data to reach the full potential of personalized medicine in type 2 diabetes. In this study, we developed a novel data-driven prescriptive algorithm for type 2 diabetes that improves significantly on the standard of care when tested on patient-level EMR data from a large medical center.

Our work is a key step toward a fully patient-centered approach to diabetes management. The authors thank Dr. Michael Kane, Massachusetts Institute of Technology, for sharing clinical expertise in the progression and treatment of diabetes and Dr.

William Adams, Boston University Clinical and Translational Science Institute, for sharing clinical expertise and assisting with the interpretation of EMR.

Savitha Prsonalized, MD, is carr assistant professor of medicine, and Irl B. Personalized diabetes care, MD, is professor of medicine Personalized diabetes care Perrsonalized Division of Metabolism, Endocrinology, and Nutrition at the University of Washington in Seattle. Savitha SubramanianIrl B. Hirsch; Personalized Diabetes Management: Moving from Algorithmic to Individualized Therapy. Diabetes Spectr 1 May ; 27 2 : 87— Personalized diabetes care

Author: Kesar

2 thoughts on “Personalized diabetes care

Leave a comment

Yours email will be published. Important fields a marked *

Design by ThemesDNA.com