One of the areas where Artificial Intelligence is having more impact is machine learning, which develops algorithms able to learn patterns and decision rules from data. Machine learning algorithms have been embedded into data mining pipelines, which can combine them with classical statistical strategies, to extract knowledge from data. Within the EU-funded MOSAIC project, a data mining pipeline has been used to derive a set of predictive models of type 2 diabetes mellitus (T2DM) complications based on electronic health record data of nearly one thousand patients. Such pipeline comprises clinical center profiling, predictive model targeting, predictive model construction and model validation. After having dealt with missing data by means of random forest (RF) and having applied suitable strategies to handle class imbalance, we have used Logistic Regression with stepwise feature selection to predict the onset of retinopathy, neuropathy, or nephropathy, at different time scenarios, at 3, 5, and 7 years from the first visit at the Hospital Center for Diabetes (not from the diagnosis). Considered variables are gender, age, time from diagnosis, body mass index (BMI), glycated hemoglobin (HbA1c), hypertension, and smoking habit. Final models, tailored in accordance with the complications, provided an accuracy up to 0.838. Different variables were selected for each complication and time scenario, leading to specialized models easy to translate to the clinical practice.

Faglia E,

Our study demonstrates that decision support tools based on the integration of multiple-source data and visual and predictive analytics do improve the management of a chronic disease such as type 2 diabetes by enacting a successful implementation of the learning health care system cycle.

SummaryObjective Metformin is widely used for the treatment of type 2 diabetes. Growing evidence supports the beneficial effects of metformin also in patients with polycystic ovary syndrome (PCOS). It was recently reported that metformin has a TSHlowering effect in hypothyroid patients with diabetes being treated with metformin. Design Aim of this study was to evaluate the effect of metformin treatment on the thyroid hormone profile in patients with PCOS. Patients and measurements Thirty-three patients with PCOS were specifically selected for being either treated with levothyroxine for a previous diagnosis of hypothyroidism (n = 7), untreated subclinically hypothyroid (n = 2) or euthyroid without levothyroxine treatment (n = 24) before the starting of metformin. The serum levels of TSH and FT 4 were measured before and after a 4-month period of metformin therapy. Results Thyroid function parameters did not change after starting metformin therapy in euthyroid patients with PCOS. In the 9 hypothyroid patients with PCOS, the basal median serum levels of TSH (3AE2 mIU/l, range = 0AE4-7AE1 mIU/l) significantly (P < 0AE05) decreased after a 4-month course of metformin treatment (1AE7 mIU/l, range = 0AE5-5AE2 mIU/l). No significant change in the serum levels of FT4 was observed in these patients. The TSH-lowering effect of metformin was not related to the administered dose of the drug, which was similar in euthyroid as compared with hypothyroid patients with PCOS (1406 ± 589 vs 1322 ± 402 mg/day, respectively; NS). Conclusions These results indicate that metformin treatment has a TSH-lowering effect in hypothyroid patients with PCOS, both treated with l-thyroxine and untreated.

OBJECTIVES:The primary aim of this study was to investigate the value of attenuation imaging (ATI), a novel ultrasound technique for detection of steatosis, by comparing the results to that obtained with controlled attenuation parameter (CAP) and by using MRI-derived proton density fat fraction (PDFF) as reference standard.METHODS:From March to November 2018, 114 consecutive adult subjects potentially at risk of steatosis and 15 healthy controls were enrolled. Each subject underwent ATI and CAP assessment on the same day. MRI-PDFF was performed within a week.RESULTS:The prevalence of steatosis, as defined by MRI-PDFF ≥ 5%, was 70.7%. There was a high correlation of ATI with MRI-PDFF (r = 0.81, P < 0.0001). The correlation of CAP with MRI-PDFF and with ATI, respectively, was moderate (r = 0.65, P < 0.0001 and r = 0.61, P < 0.0001). The correlation of ATI or CAP with PDFF was not affected by age, gender, or body mass index. Area under the receiver operating characteristics of ATI and CAP, respectively, were 0.91 (0.84–0.95; P < 0.0001) and 0.85 (0.77–0.91; P < 0.0001) for detecting S > 0 steatosis (MRI-PDFF ≥ 5%); 0.95 (0.89–0.98; P < 0.0001) and 0.88 (0.81–0.93; P < 0.0001) for detecting S > 1 steatosis (MRI-PDFF ≥ 16.3%). The cutoffs of ATI and CAP, respectively, were 0.63 dB/cm/MHz and 258 dB/m for detecting S > 0 liver steatosis; 0.72 dB/cm/MHz and 304 dB/m for detecting S > 1 steatosis. ATI performed better than CAP, and this improvement was statistically significant for S > 1 (P = 0.04).DISCUSSION:This study shows that, in patients with no fibrosis/mild fibrosis, ATI is a very promising tool for the noninvasive assessment of steatosis.