A transparent machine learning algorithm uncovers HbA1c patterns associated with therapeutic inertia in patients with type 2 diabetes and failure of metformin monotherapy.

AIMS: This study aimed to identify and categorize the determinants influencing the intensification of therapy in Type 2 Diabetes (T2D) patients with suboptimal blood glucose control despite metformin monotherapy. METHODS: Employing the Logic Learning Machine (LLM), an advanced artificial intelligence system, we scrutinized electronic health records of 1.5 million patients treated in 271 diabetes clinics affiliated with the Italian Association of Medical Diabetologists from 2005 to 2019. Inclusion criteria comprised patients on metformin monotherapy with two consecutive mean HbA1c levels exceeding 7.0%. The cohort was divided into "inertia-NO" (20,067 patients with prompt intensification) and "inertia-YES" (13,029 patients without timely intensification). RESULTS: The LLM model demonstrated robust discriminatory ability among the two groups (ROC-AUC = 0.81, accuracy = 0.71, precision = 0.80, recall = 0.71, F1 score = 0.75). The main novelty of our results is indeed the identification of two main distinct subtypes of therapeutic inertia. The first exhibited a gradual but steady HbA1c increase, while the second featured a moderate, non-uniform rise with substantial fluctuations. CONCLUSIONS: Our analysis sheds light on the significant impact of HbA1c levels over time on therapeutic inertia in patients with T2D, emphasizing the importance of early intervention in the presence of specific HbA1c patterns.

Achieving Good Metabolic Control Without Weight Gain with the Systematic Use of GLP-1-RAs and SGLT-2 Inhibitors in Type 2 Diabetes: A Machine-learning Projection Using Data from Clinical Practice.

PURPOSE: Recently, the 2022 American Diabetes Association and European Association for the Study of Diabetes (ADA-EASD) consensus report stressed the importance of weight control in the management of patients with type 2 diabetes; weight control should be a primary target of therapy. This retrospective analysis evaluated, through an artificial-intelligence (AI) projection of data from the AMD Annals database-a huge collection of most Italian diabetology medical records covering 15 years (2005-2019)-the potential effects of the extended use of sodium-glucose co-transporter 2 inhibitors (SGLT-2is) and of glucose-like peptide 1 receptor antagonists (GLP-1-RAs) on HbA1c and weight. METHODS: Data from 4,927,548 visits in 558,097 patients were retrospectively extracted using these exclusion criteria: type 1 diabetes, pregnancy, age >75 years, dialysis, and lack of data on HbA1c or weight. The analysis revealed late prescribing of SGLT-2is and GLP-1-RAs (innovative drugs), and considering a time frame of 4 years (2014-2017), a paradoxic greater percentage of combined-goal (HbA1c <7% and weight gain <2%) achievement was found with older drugs than with innovative drugs, demonstrating aspects of therapeutic inertia. Through a machine-learning AI technique, a "what-if" analysis was performed, using query models of two outcomes: (1) achievement of the combined goal at the visit subsequent to a hypothetical initial prescribing of an SGLT-2i or a GLP-1-RA, with and without insulin, selected according to the 2018 ADA-EASD diabetes recommendations; and (2) persistence of the combined goal for 18 months. The precision values of the two models were, respectively, sensitivity, 71.1 % and 69.8%, and specificity, 67% and 76%. FINDINGS: The first query of the AI analysis showed a great improvement in achievement of the combined goal: 38.8% with prescribing in clinical practice versus 66.5% with prescribing in the "what-if" simulation. Addressing persistence at 18 months after the initial achievement of the combined goal, the simulation showed a potential better performance of SGLT-2is and GLP-1-RAs with respect to each antidiabetic pharmacologic class or combination considered. IMPLICATIONS: AI appears potentially useful in the analysis of a great amount of data, such as that derived from the AMD Annals. In the present study, an LLM analysis revealed a great potential improvement in achieving metabolic targets with SGLT-2i and GLP-1-RA utilization. These results underscore the importance of early, timely, and extended use of these new drugs.

Artificial Intelligence and Big Data in Diabetes Care: A Position Statement of the Italian Association of Medical Diabetologists.

Since the last decade, most of our daily activities have become digital. Digital health takes into account the ever-increasing synergy between advanced medical technologies, innovation, and digital communication. Thanks to machine learning, we are not limited anymore to a descriptive analysis of the data, as we can obtain greater value by identifying and predicting patterns resulting from inductive reasoning. Machine learning software programs that disclose the reasoning behind a prediction allow for "what-if" models by which it is possible to understand if and how, by changing certain factors, one may improve the outcomes, thereby identifying the optimal behavior. Currently, diabetes care is facing several challenges: the decreasing number of diabetologists, the increasing number of patients, the reduced time allowed for medical visits, the growing complexity of the disease both from the standpoints of clinical and patient care, the difficulty of achieving the relevant clinical targets, the growing burden of disease management for both the health care professional and the patient, and the health care accessibility and sustainability. In this context, new digital technologies and the use of artificial intelligence are certainly a great opportunity. Herein, we report the results of a careful analysis of the current literature and represent the vision of the Italian Association of Medical Diabetologists (AMD) on this controversial topic that, if well used, may be the key for a great scientific innovation. AMD believes that the use of artificial intelligence will enable the conversion of data (descriptive) into knowledge of the factors that "affect" the behavior and correlations (predictive), thereby identifying the key aspects that may establish an improvement of the expected results (prescriptive). Artificial intelligence can therefore become a tool of great technical support to help diabetologists become fully responsible of the individual patient, thereby assuring customized and precise medicine. This, in turn, will allow for comprehensive therapies to be built in accordance with the evidence criteria that should always be the ground for any therapeutic choice.

Impact of severe and symptomatic hypoglycemia on quality of life and fear of hypoglycemia in type 1 and type 2 diabetes. Results of the Hypos-1 observational study.

BACKGROUND AND AIMS: Hypoglycemia represents a relevant burden in people with diabetes. Consequences of hypoglycemia/fear of hypoglycemia on quality of life (QoL) and behaviors of patients with T1DM and T2DM were assessed. METHODS AND RESULTS: HYPOS-1 was an observational retrospective study. Fear of hypoglycemia (Fear of Hypoglycemia Questionnaire, FHQ), general health status (visual analog scale of EuroQol questionnaire, EQ5D-VAS) psychological well-being (WHO-5 well being index, WHO-5), diabetes related distress (Problem Areas in Diabetes 5, PAID-5), and corrective/preventive behaviors following hypoglycemia were compared between people with and without previous experience of severe and symptomatic hypoglycemia and by tertiles of FHQ scores. A multivariate analysis was performed to identify factors associated with the likelihood of being in the third tertile of FHQ score. Overall, 2229 patients were involved. Severe hypoglycemia had statistically significant and clinically relevant (measured as effect sizes) negative impact on EQ5D-VAS, WHO-5, PAID-5, and FHQ both in T1DM and T2DM. In T2DM, symptomatic episodes had similar impact of severe hypoglycemia. Moving from the first to the third FHQ tertile, lower scores of EQ-5D VAS and WHO-5, and higher levels of PAID-5 were found. Patients in the third tertile performed more frequently corrective/preventive actions that negatively impact on metabolic control. Previous hypoglycemia, insulin treatment, female gender, age, and school education were the independent factors associated with increased likelihood to be in the third tertile. CONCLUSION: Not only severe but also symptomatic hypoglycemia negatively affect patient QoL, especially in T2DM. Addressing fear of hypoglycemia should be a goal of diabetes education.

Real-world use of self-monitoring of blood glucose in people with type 2 diabetes: an urgent need for improvement.

AIMS: To assess use of self-monitoring of blood glucose (SMBG) in type 2 diabetes (T2DM) in the context of a continuous quality improvement initiative (AMD Annals). METHODS: 14 quality-of-care indicators were developed, including frequency of SMBG, fasting blood glucose (FBG), and post-prandial glucose (PPG) levels, and hypoglycemia and hyperglycemia episodes. Clinical data and SMBG values downloaded from any glucose meter were obtained from electronic medical records. The most frequently used glucose-lowering treatment regimens were identified and the indicators were assessed separately by regimen. RESULTS: Overall, 21 Italian centers and 13,331 patients (accounting for 35,657 HbA1c tests and 8.44 million SMBG values collected during 2014 and 2015) were included in the analysis; 11 therapeutic regimens were selected. Patients in regimens not including insulin performed 15-23 measurements per patient-month, those treated with basal insulin 32.1 tests/patient-month, and those treated with basal and short-acting insulin 53-58 tests/patient-month. In all treatment regimens, PPG measurements represented a minority of all tests; pre-breakfast measurements accounted for about 50% of all FBG values. Mean FBG levels exceeded 130 mg/dl in 49.3-88.3% of the cases in the different treatment regimens, while PPG levels were over 140 mg/dl in 46.7-81.0%. From 5.7 to 32.7%, patients in the different regimens had at least one episode of hypoglycemia (< 70 mg/dl), while from 3.7 to 47.7% had at least one episode of hyperglycemia (> 300 mg/dl). CONCLUSIONS: SMBG is underutilized in patients with T2DM treated or not with insulin. In all treatment groups, PPG is seldom investigated. Poor metabolic control and rates of hyper- and hypoglycemia deserve consideration in all treatment groups.

Incidence and risk factors for severe and symptomatic hypoglycemia in type 1 diabetes. Results of the HYPOS-1 study.

OBJECTIVE: Hypoglycemia is common in type 1 diabetes mellitus (T1DM). We aimed to update the incidence of severe and symptomatic hypoglycemia and investigate several correlated factors. METHODS: In this multicenter, observational retrospective study, the data of 206 T1DM patients from a sample of 2,229 consecutive patients seen at 18 diabetes clinics were analyzed. Sociodemographic and clinical characteristics, severe hypoglycemia in the past 12 months, and symptomatic hypoglycemia in the past 4 weeks were recorded with a self-report questionnaire and a clinical form during a routine visit. Poisson multivariate models were applied. RESULTS: A minority of patients accounted for the majority of both severe and symptomatic episodes. The incidence rate (IR) of severe hypoglycemia was 0.49 (0.40-0.60) events/person-years. The incidence rate ratio (IRR) was higher in patients with previous severe hypoglycemia (3.71; 2.28-6.04), neuropathy (4.16; 2.14-8.05), long duration (>20 years, 2.96; 1.60-5.45), and on polypharmacy (1.24; 1.13-1.36), but it was lower when a complication was present. The IR of symptomatic hypoglycemia was 53.3 events/person-years, with an IRR significantly higher among women or patients with better education, or shorter duration or on pumps. The IRR was lower in patients with higher BMI or neuropathy or aged more than 50 years. CONCLUSIONS: Fewer than 20 % of T1DM patients are free from hypoglycemia, with one in six having experienced at least one severe episode in the last year. The distribution is uneven, with a tendency of episodes to cluster in some patients. Severe and symptomatic episodes have different correlates and reflect different conditions.

Self-monitoring of blood glucose: guideline application rather than utilization restrictions on testing strips has potential to reduce diabetes healthcare costs in Italy.

BACKGROUND: Self-monitoring of blood glucose (SMBG) is a proven tool to improve glycemic control, even if it might increase direct costs for diabetes management. In Italy, the purchase, prescription rules and responsibilities, and distribution of testing strips per type of patient are managed differently in each of the 20 Italian regions. The Italian scientific societies for diabetes (Società Italiana Diabetologia [SID] and Associazione Medici Diabetologi [AMD]) have issued validated guidelines for SMBG, but not all regions apply them. We investigated whether following SID-AMD guidelines would help decreasing SMBG and diabetes healthcare costs in Italy. MATERIALS AND METHODS: We compared the regions applying and not applying SMBG guidelines for the mean number of testing strips used, number of hospitalizations (with the principal diagnosis of diabetes, excluding diabetes complications), and duration of hospitalization, as indirect measures of SMBG cost. RESULTS: Regions applying the guidelines recorded higher SMBG testing strip utilization than regions not applying guidelines, but they recorded fewer hospitalizations for diabetes (36.2 ± 11.3 vs. 79.9 ± 27.8 hospitalizations per 100,000 inhabitants, P<0.002) and fewer days in the hospital (363 ± 106 vs. 685 ± 194 days of hospitalization for diabetes per 100,000 inhabitants, P<0.002). CONCLUSIONS: Our data suggest that application of guidelines for SMBG prescription and a strict cooperation between health providers and regional health economic deciders were associated with greater utilization of SMBG testing strips. They were also associated with significantly reduced number of hospitalizations and reduced overall duration of hospitalization for patients with diabetes, potentially saving healthcare costs.