Adaptive Algorithms for Personalized Diabetes Treatment

Dynamic systems, especially in real-life applications, are often determined by inter-/intra-variability, uncertainties and time-varying components. Physiological systems are probably the most representative example in which population variability, vital signal measurement noise and uncertain dynamics render their explicit representation and optimization a rather difficult task. Systems characterized by such challenges often require the use of adaptive algorithmic solutions able to perform an iterative structural and/or parametrical update process towards optimized behavior. Adaptive optimization presents the advantages of (i) individualization through learning of basic system characteristics, (ii) ability to follow time-varying dynamics and (iii) low computational cost. In this chapter, the use of online adaptive algorithms is investigated in two basic research areas related to diabetes management: (i) real-time glucose regulation and (ii) real-time prediction of hypo-/hyperglycemia. The applicability of these methods is illustrated through the design and development of an adaptive glucose control algorithm based on reinforcement learning and optimal control and an adaptive, personalized early-warning system for the recognition and alarm generation against hypo- and hyperglycemic events.

Personalized Tuning of a Reinforcement Learning Control Algorithm for Glucose Regulation

Artificial pancreas is in the forefront of research towards the automatic insulin infusion for patients with type 1 diabetes. Due to the high inter- and intra-variability of the diabetic population, the need for personalized approaches has been raised. This study presents an adaptive, patient-specific control strategy for glucose regulation based on reinforcement learning and more specifically on the Actor-Critic (AC) learning approach. The control algorithm provides daily updates of the basal rate and insulin-to-carbohydrate (IC) ratio in order to optimize glucose regulation. A method for the automatic and personalized initialization of the control algorithm is designed based on the estimation of the transfer entropy (TE) between insulin and glucose signals. The algorithm has been evaluated in silico in adults, adolescents and children for 10 days. Three scenarios of initialization to i) zero values, ii) random values and iii) TE-based values have been comparatively assessed. The results have shown that when the TE-based initialization is used, the algorithm achieves faster learning with 98%, 90% and 73% in the A+B zones of the Control Variability Grid Analysis for adults, adolescents and children respectively after five days compared to 95%, 78%, 41% for random initialization and 93%, 88%, 41% for zero initial values. Furthermore, in the case of children, the daily Low Blood Glucose Index reduces much faster when the TE-based tuning is applied. The results imply that automatic and personalized tuning based on TE reduces the learning period and improves the overall performance of the AC algorithm.

Early metacognitive abilities: The interplay of monitoring and control processes in 5- to 7-year-old children

The goal of the current investigation was to compare two monitoring processes (judgments of learning [JOLs] and confidence judgments [CJs]) and their corresponding control processes (allocation of study time and selection of answers to maximize accuracy, respectively) in 5- to 7-year-old children (N=101). Children learned the meaning of Japanese characters and provided JOLs after a study phase and CJs after a memory test. They were given the opportunity to control their learning in self-paced study phases, and to control their accuracy by placing correct answers into a treasure chest and incorrect answers into a trash can. All three age groups gave significantly higher CJs for correct compared to incorrect answers, with no age-related differences in the magnitude of this difference, suggesting robust metacognitive monitoring skills in children as young as 5. Furthermore, a link between JOLs and study time was found in the 6- and 7-year-olds, such that children spent more time studying items with low JOLs compared to items with high JOLs. Also, 6- and 7-year-olds but not 5-year-olds spent more time studying difficult items compared to easier items. Moreover, age-related improvements were found in children's use of CJs to guide their selection of answers: although children as young as 5 placed their most confident answers in the treasure chest and least confident answers in the trash can, this pattern was more robust in older children. Overall, results support the view that some metacognitive judgments may be acted upon with greater ease than others among young children.

The impact of retrieval processes, age, general achievement level, and test scoring scheme for children's metacognitive monitoring and controlling

This multi-phase study examined the influence of retrieval processes on children’s metacognitive processes in relation to and in interaction with achievement level and age. First, N = 150 9/10- and 11/12-year old high and low achievers watched an educational film and predicted their test performance. Children then solved a cloze test regarding the film content including answerable and unanswerable items and gave confidence judgments to every answer. Finally, children withdrew answers that they believed to be incorrect. All children showed adequate metacognitive processes before and during test taking with 11/12- year-olds outperforming 9/10-year-olds when considering characteristics of on-going retrieval processes. As to the influence of achievement level, high compared to low achievers proved to be more accurate in their metacognitive monitoring and controlling. Results suggest that both cognitive resources (operationalized through achievement level) and mnemonic experience (assessed through age) fuel metacognitive development. Nevertheless, when facing higher demands regarding retrieval processes, experience seems to play the more important role.

An indicator framework for assessing livelihood resilience in the context of social–ecological dynamics

Livelihood resilience draws attention to the factors and processes that keep livelihoods functioning despite change and thus enriches the livelihood approach which puts people, their differential capabilities to cope with shocks and how to reduce poverty and improve adaptive capacity at the centre of analysis. However, the few studies addressing resilience from a livelihood perspective take different approaches and focus only on some dimensions of livelihoods. This paper presents a framework that can be used for a comprehensive empirical analysis of livelihood resilience. We use a concept of resilience that considers agency as well as structure. A review of both theoretical and empirical literature related to livelihoods and resilience served as the basis to integrate the perspectives. The paper identifies the attributes and indicators of the three dimensions of resilience, namely, buffer capacity, self-organisation and capacity for learning. The framework has not yet been systematically tested; however, potentials and limitations of the components of the framework are explored and discussed by drawing on empirical examples from literature on farming systems. Besides providing a basis for applying the resilience concept in livelihood-oriented research, the framework offers a way to communicate with practitioners on identifying and improving the factors that build resilience. It can thus serve as a tool for monitoring the effectiveness of policies and practices aimed at building livelihood resilience.