Personalizing Self-management in Diabetes - Pilot Study

Dynamically Tailoring Interventions for Problem-Solving in Diabetes Self-Management Using Self-Monitoring Data

The goal of this study is to conduct a pilot feasibility study a novel informatics intervention, GlucoType (also called Platano for Latino users) that incorporates computational analysis of self-monitoring data to help individuals with type 2 diabetes personalize diabetes self-management strategies. This study will include 20 individuals with type 2 diabetes mellitus (T2DM) recruited from economically disadvantaged and medically underserved communities to test Platano for 4 weeks to assess its acceptability and feasibility. The main outcome measures include problem-solving abilities in diabetes (Diabetes Problem-Solving Inventory (DPSA)) and self-reported diabetes self-care (Summary of Diabetes Self-Care Activities Questionnaire (SDSCA)). In addition, this study will include a controlled laboratory experiment to assess whether participants can understand and follow personalized nutritional goals generated by Platano.

Growing evidence highlights significant differences in individuals' physiology and glycemic function and their cultural, social, and economical circumstances that impact diabetes self-management. These discoveries paved the way for precision medicine-an approach to personalizing medical treatment to an individual's genetic makeup, clinical history, and lifestyle. Computational learning methods have been successfully used for identifying clinical phenotypes-observable manifestations of diseases. Studies showed the benefits of tailoring not only medical treatment, but also behavioral interventions; however, tailoring typically relies on expert identification of tailoring variables and decision rules, and on standard surveys. Data collected with self-monitoring can more accurately reflect an individual's behaviors and glycemic patterns, thus highlighting their "behavioral phenotypes", yet such data are rarely utilized in tailoring. The ongoing focus of this research is on facilitating problem-solving in diabetes self-management. Well-developed problem-solving skills are essential to diabetes management result in better diabetes self-care behaviors lead to improvements in clinical outcomes and can be fostered with face-to-face interventions. Previous research suggested problem identification and generation of alternatives as critical steps in problem-solving in diabetes. In previous work, the investigators developed an informatics intervention that relied on expert-generated knowledge for assisting individuals on these steps of problem-solving. In this pilot feasibility study, the investigators study an alternative solution that relies on computational pattern analysis of data collected with self-monitoring technologies to tailor the problem-solving assistance to individuals' unique behavioral phenotypes. The intervention, GlucoType uses computational learning methods to identify systematic patterns in individuals' diet, physical activity, and sleep, captured with custom-built and commercial self-monitoring technologies, and correlates these patterns with fluctuations in individuals' blood glucose levels. GlucoType then uses this information to 1) identify behavioral patterns associated with high glycemic excursion, 2) formulate personalized goals to modify these behaviors, 3) provide in-the-moment decision support to help individuals be more consistent in meeting their goals.

GlucoType is an mobile Health intervention for facilitating self-management in T2DM built for iPhone and Android smartphones. GlucoType includes a custom-built interface for low-burden capture of diet and blood glucose (BG) levels and relies on a commercial activity tracker, FitBit, for capture of sleep and physical activity. It then applies computational phenotyping techniques to identify patterns of associations between daily activities and changes in BG levels. GlucoType uses an expert system developed by our research team to translate identified phenotypes into automatically-generated personalized behavioral goals for improving glycemic control formulated in natural language.

Change in score on Summary of Diabetes Self-Care Activities Questionnaire (SDSCA) - 12-item with 5 sub-scales (diet, exercise, home blood glucose testing, foot care, smoking status). The respondent is asked how many days in the past week he/she performed the behavior (ranges from 0 to 7); higher scores indicates higher performance.

Inclusion Criteria: Age 18-65 years A diagnosis of Type 2 Diabetes. A participant of the Washington Heights/Inwood Informatics Infrastructure for Comparative Effectiveness Research (WICER), a patient of the AIM clinic, or a patient of a participating Federally Qualified Health Center (FQHC) health center for at least 6 months Has participated in at least one diabetes education session at the participating site in the last 6 months Proficient in either English or Spanish Must own a basic cell phone Exclusion Criteria: Pregnancy Presence of serious illness (e.g. cancer diagnosis with active treatment, advanced stage heart failure, multiple sclerosis) Presence of cognitive impairment Plans for leaving their healthcare provider in the next 12 months Does not have a computer and/or Internet access

Zeevi D, Korem T, Zmora N, Israeli D, Rothschild D, Weinberger A, Ben-Yacov O, Lador D, Avnit-Sagi T, Lotan-Pompan M, Suez J, Mahdi JA, Matot E, Malka G, Kosower N, Rein M, Zilberman-Schapira G, Dohnalova L, Pevsner-Fischer M, Bikovsky R, Halpern Z, Elinav E, Segal E. Personalized Nutrition by Prediction of Glycemic Responses. Cell. 2015 Nov 19;163(5):1079-1094. doi: 10.1016/j.cell.2015.11.001.

Haas L, Maryniuk M, Beck J, Cox CE, Duker P, Edwards L, Fisher EB, Hanson L, Kent D, Kolb L, McLaughlin S, Orzeck E, Piette JD, Rhinehart AS, Rothman R, Sklaroff S, Tomky D, Youssef G; 2012 Standards Revision Task Force. National standards for diabetes self-management education and support. Diabetes Care. 2013 Jan;36 Suppl 1(Suppl 1):S100-8. doi: 10.2337/dc13-S100. No abstract available.

Collins FS, Varmus H. A new initiative on precision medicine. N Engl J Med. 2015 Feb 26;372(9):793-5. doi: 10.1056/NEJMp1500523. Epub 2015 Jan 30.

Hastie T, Tibshirani R, Friedman J. The Elements of Statistical Learning [Internet]. New York, NY: Springer New York; 2009 [cited 2016 Jun 4]. (Springer Series in Statistics)

Liao KP, Cai T, Savova GK, Murphy SN, Karlson EW, Ananthakrishnan AN, Gainer VS, Shaw SY, Xia Z, Szolovits P, Churchill S, Kohane I. Development of phenotype algorithms using electronic medical records and incorporating natural language processing. BMJ. 2015 Apr 24;350:h1885. doi: 10.1136/bmj.h1885.

Hripcsak G, Albers DJ. Next-generation phenotyping of electronic health records. J Am Med Inform Assoc. 2013 Jan 1;20(1):117-21. doi: 10.1136/amiajnl-2012-001145. Epub 2012 Sep 6.