Juan Lanchares

Modern AI applications contain computationally expensive sections. Accelerator cards and tools like AMD Vitis HLS leverage high-level synthesis (HLS) and hardware (HW) optimizations to create custom HW designs to accelerate them. Nevertheless, the learning curve is steep, even for those with previous knowledge of HW design, due to the complexity of the optimization techniques and limited information on their interactions and HW effects. This article quantitatively analyses the interactions of optimization techniques after reverse engineering Vitis’ optimization directives, both in isolation and in pairs. Over 150 experiments were conducted to investigate three distinct goals: 1) assessing pragma behavior and the rules governing pragma application and optimizations; 2) modeling Vitis HLS latency estimates; and 3) evaluating the impact of optimizations on design space exploration (DSE), specifically area and latency. These experiments involve different combinations and placements of optimizations in the loop and function hierarchy of the test bench. Our findings offer guidance on using Vitis pragmas and identify promising configurations for optimizing latency and area.

This article proposes the first hardware implemen-tation of a low-power LSTM neural network targeting a wearable medical device designed to predict blood glucose at a 30-minute horizon. This work aims to reduce energy consumption by propos-ing new activation functions that target hardware implementation. On top of this proposal, we also prove there is room for improve-ment in energy consumption by applying neural network optimiza-tions at the algorithmic, such as quantization, and architecture level, LSTM hyperparameters, that consider the target hardware. To validate our proposal, we devise an optimized version of the neural network aimed to be wearable and, therefore, to reduce its energy consumption while preserving its accuracy as much as possible. The hardware is implemented on a Xilinx Virtex-7 FPGA VC707 Evaluation Kit. It is compared with (i) a faithful design of the original neural network implemented on the same evaluation kit, (ii) three state-of-the-art LSTM-based FPGA implementations, and (iii) software implementations running in cutting-edge smartphones:OnePlus NordTM and an Apple iPhone 13 ProTM with artificial in-telligence hardware accelerators. Our proposal consumes between ×1020 and ×7 less energy than the software implementations, being the most efficient system compared to the smartphones. On the other hand, its energy efficiency, measured in GFLOP/J, is between ×2.84 and ×7.82 greater than other state-of-the-art LSTM implementations, proving to be the most suitable implementation for a wearable system for blood glucose prediction.

Objective: Assess in a sample of people with type 1 diabetes mellitus whether mood and stress influence blood glucose levels and variability. Material and Methods: Continuous glucose monitoring was performed on 10 patients with type 1 diabetes mellitus, where interstitial glucose values were recorded every 15 min. A daily survey was conducted through Google Forms, collecting information on mood and stress. The day was divided into six slots of 4-h each, asking the patient to assess each slot in relation to mood (sad, normal or happy) and stress (calm, normal or nervous). Different measures of glycemic control (arithmetic mean and percentage of time below/above the target range) and variability (standard deviation, percentage coefficient of variation, mean amplitude of glycemic excursions and mean of daily differences) were calculated to relate the mood and stress perceived by patients with blood glucose levels and glycemic variability. A hypothesis test was carried out to quantitatively compare the data groups of the different measures using the Student’s t-test. Results: Statistically significant differences (p-value < 0.05) were found between different levels of stress. In general, average glucose and variability decrease when the patient is calm. There are statistically significant differences (p-value < 0.05) between different levels of mood. Variability increases when the mood changes from sad to happy. However, the patient’s average glucose decreases as the mood improves. Conclusions: Variations in mood and stress significantly influence blood glucose levels, and glycemic variability in the patients analyzed with type 1 diabetes mellitus. Therefore, they are factors to consider for improving glycemic control. The mean of daily differences does not seem to be a good indicator for variability.

This article proposes two ensemble neural network-based models for blood glucose prediction at three different prediction horizons—30, 60, and 120 min—and compares their performance with ten recently proposed neural networks. The twelve models’ performances are evaluated under the same OhioT1DM Dataset, preprocessing workflow, and tools at the three prediction horizons using the most common metrics in blood glucose prediction, and we rank the best-performing ones using three methods devised for the statistical comparison of the performance of multiple algorithms: scmamp, model confidence set, and superior predictive ability. Our analysis provides a comparison of the state-of-the-art neural networks for blood glucose prediction, estimating the model’s error, highlighting those with the highest probability of being the best predictors, and providing a guide for their use in clinical practice.

The demand for Continuous Glucose Monitoring systems is increasing among type 1 diabetic patients. Some companies are trying to improve the monitoring and the usability of these systems. One example is Abbott FreeStyle Libre, which provides a new concept of glucose monitoring called Flash Glucose Monitoring which is more affordable and does not need calibration. The increasing demand for these devices means an opportunity for data and computer scientists, who can contribute to the development of decision-making support systems based on the data collected from the devices. Type 1 diabetic patients that use FreeStyle Libre may enter the number of insulin and carbohydrates units that they are going to take before a meal. Using both the entered data and the blood glucose values collected by the device automatically, the application presented in this paper generates a report of the patient’s glucose patterns. In addition, it provides a web application that allows the user to upload the data obtained from the device and download the report on his computer or smartphone. The application uses decision trees to detect the patterns and entails a starting point in the creation of ensemble models with more predictive power, also based on decision trees. Furthermore, the methodology makes a segmentation of the data set in blocks, determined by the different meals done throughout the day, adding more information to the set of variables used to train the model. As a result, the application can discover repetitive patterns in the daily life of the patient, which can help to take early preventive measures for risk situations in a period close to the next meal.

Objective: Assess in a sample of patients with type 1 diabetes mellitus whether mood and stress influence blood glucose levels and variability. Material and Methods: Continuous glucose monitoring was performed on 10 patients with type 1 diabetes, where interstitial glucose values were recorded every 15 minutes. A daily survey was conducted through Google Forms, collecting information on mood and stress. The day was divided into 6 slots of 4-hour each, asking the patient to assess each slot in relation to mood (sad, normal or happy) and stress (calm, normal or nervous). Different measures of glycemic control (arithmetic mean and percentage of time below/above the target range) and variability (standard deviation, percentage coefficient of variation, mean amplitude of glycemic excursions and mean of daily differences) were calculated to relate the mood and stress perceived by patients with blood glucose levels and glycemic variability. A hypothesis test was carried out to quantitatively compare the data groups of the different measures using the Student’s t-test. Results: Statistically significant differences (p-value < 0.05) were found between different levels of stress. In general, average glucose and variability decrease when the patient is calm. There are statistically significant differences (p-value < 0.05) between different levels of mood. Variability increases when the mood changes from sad to happy. However, the patient’s average glucose decreases as the mood improves. Conclusions: Variations in mood and stress significantly influence blood glucose levels, and glycemic variability in the patients analyzed with type 1 diabetes mellitus. Therefore, they are factors to consider for improving glycemic control. The mean of daily differences does not seem to be a good indicator for variability.

Predicting glucose values on the basis of insulin and food intakes is a difficult task that people with diabetes need to do daily. This is necessary as it is important to maintain glucose levels at appropriate values to avoid not only short-term, but also long-term complications of the illness. Artificial intelligence in general and machine learning techniques in particular have already lead to promising results in modeling and predicting glucose concentrations. In this work, several machine learning techniques are used for the modeling and prediction of glucose concentrations using as inputs the values measured by a continuous monitoring glucose system as well as also previous and estimated future carbohydrate intakes and insulin injections. In particular, we use the following four techniques: genetic programming, random forests, k-nearest neighbors, and grammatical evolution. We propose two new enhanced modeling algorithms for glucose prediction, namely (i) a variant of grammatical evolution which uses an optimized grammar, and (ii) a variant of tree-based genetic programming which uses a three-compartment model for carbohydrate and insulin dynamics. The predictors were trained and tested using data of ten patients from a public hospital in Spain. We analyze our experimental results using the Clarke error grid metric and see that 90% of the forecasts are correct (i.e., Clarke error categories A and B), but still even the best methods produce 5 to 10% of serious errors (category D) and approximately 0.5% of very serious errors (category E). We also propose an enhanced genetic programming algorithm that incorporates a three-compartment model into symbolic regression models to create smoothed time series of the original carbohydrate and insulin time series.