Distributed learning: a reliable privacy-preserving strategy to change multicenter collaborations using AI.

PURPOSE: The present scoping review aims to assess the non-inferiority of distributed learning over centrally and locally trained machine learning (ML) models in medical applications. METHODS: We performed a literature search using the term "distributed learning" OR "federated learning" in the PubMed/MEDLINE and EMBASE databases. No start date limit was used, and the search was extended until July 21, 2020. We excluded articles outside the field of interest; guidelines or expert opinion, review articles and meta-analyses, editorials, letters or commentaries, and conference abstracts; articles not in the English language; and studies not using medical data. Selected studies were classified and analysed according to their aim(s). RESULTS: We included 26 papers aimed at predicting one or more outcomes: namely risk, diagnosis, prognosis, and treatment side effect/adverse drug reaction. Distributed learning was compared to centralized or localized training in 21/26 and 14/26 selected papers, respectively. Regardless of the aim, the type of input, the method, and the classifier, distributed learning performed close to centralized training, but two experiments focused on diagnosis. In all but 2 cases, distributed learning outperformed locally trained models. CONCLUSION: Distributed learning resulted in a reliable strategy for model development; indeed, it performed equally to models trained on centralized datasets. Sensitive data can get preserved since they are not shared for model development. Distributed learning constitutes a promising solution for ML-based research and practice since large, diverse datasets are crucial for success.

A virtual reality classroom to teach and explore crystal solid state structures.

We present an educational application of virtual reality that we created to help students gain an in-depth understanding of the internal structure of crystals and related key concepts. Teachers can use it to give lectures to small groups (10-15) of students in a shared virtual environment, both remotely (with teacher and students in different locations) and locally (while sharing the same physical space). Lectures can be recorded, stored in an online repository, and shared with students who can either review a recorded lecture in the same virtual environment or can use the application for self-studying by exploring a large collection of available crystal structures. We validated our application with human subjects receiving positive feedback. Supplementary Information: The online version contains supplementary material available at 10.1007/s11042-022-13410-0https://doi.org/10.1007/s11042-022-13410-0.

Improving reading skills in students with dyslexia: the efficacy of a sublexical training with rhythmic background.

The core deficit underlying developmental dyslexia (DD) has been identified in difficulties in dynamic and rapidly changing auditory information processing, which contribute to the development of impaired phonological representations for words. It has been argued that enhancing basic musical rhythm perception skills in children with DD may have a positive effect on reading abilities because music and language share common mechanisms and thus transfer effects from the former to the latter are expected to occur. A computer-assisted training, called Rhythmic Reading Training (RRT), was designed in which reading exercises are combined with rhythm background. Fourteen junior high school students with DD took part to 9 biweekly individual sessions of 30 min in which RRT was implemented. Reading improvements after the intervention period were compared with ones of a matched control group of 14 students with DD who received no intervention. Results indicated that RRT had a positive effect on both reading speed and accuracy and significant effects were found on short pseudo-words reading speed, long pseudo-words reading speed, high frequency long words reading accuracy, and text reading accuracy. No difference in rhythm perception between the intervention and control group were found. Findings suggest that rhythm facilitates the development of reading skill because of the temporal structure it imposes to word decoding.

Computational Intelligence and Game Design for Effective At-Home Stroke Rehabilitation.

OBJECTIVE: The aim of this article is to describe a game engine that has all the characteristics needed to support rehabilitation at home. The low-cost tracking devices recently introduced in the entertainment market allow measuring reliably at home, in real time, players' motion with a hands-free approach. Such systems have also become a source of inspiration for researchers working in rehabilitation. Computer games appear suited to guide rehabilitation because of their ability to engage the users. However, commercial videogames and game engines lack the peculiar functionalities required in rehabilitation: Games should be adapted to each patient's functional status, and monitoring the patient's motion is mandatory to avoid maladaptation. Feedback on performance and progression of the exercises should be provided. Lastly, several tracking devices should be considered, according to the patient's pathology and rehabilitation aims. SUBJECTS AND METHODS: We have analyzed the needs of the clinicians and of the patients associated in performing rehabilitation at home, identifying the characteristics that the game engine should have. RESULTS: The result of this analysis has led us to develop the Intelligent Game Engine for Rehabilitation (IGER) system, which combines the principles upon which commercial games are designed with the needs of rehabilitation. IGER is heavily based on computational intelligence: Adaptation of the difficulty level of the exercise is carried out through a Bayesian framework from the observation of the patient's success rate. Monitoring is implemented in fuzzy systems and based on rules defined for the exercises by clinicians. Several devices can be attached to IGER through an input abstraction layer, like the Nintendo® (Kyoto, Japan) Wii™ Balance Board™, the Microsoft® (Redmond, WA) Kinect, the Falcon from Novint Technologies (Albuquerque, NM), or the Tyromotion (Graz, Austria) Timo® plate balance board. IGER is complemented with videogames embedded in a specific taxonomy developed to support rehabilitation progression through time. CONCLUSIONS: A few games aimed at postural rehabilitation have been designed and developed to test the functionalities of the IGER system. The preliminary results of tests on normal elderly people and patients with the supervision of clinicians have shown that the IGER system indeed does feature the characteristics required to support rehabilitation at home and that it is ready for clinical pilot testing at patients' homes.

Generalization in the XCSF classifier system: analysis, improvement, and extension.

We analyze generalization in XCSF and introduce three improvements. We begin by showing that the types of generalizations evolved by XCSF can be influenced by the input range. To explain these results we present a theoretical analysis of the convergence of classifier weights in XCSF which highlights a broader issue. In XCSF, because of the mathematical properties of the Widrow-Hoff update, the convergence of classifier weights in a given subspace can be slow when the spread of the eigenvalues of the autocorrelation matrix associated with each classifier is large. As a major consequence, the system's accuracy pressure may act before classifier weights are adequately updated, so that XCSF may evolve piecewise constant approximations, instead of the intended, and more efficient, piecewise linear ones. We propose three different ways to update classifier weights in XCSF so as to increase the generalization capabilities of XCSF: one based on a condition-based normalization of the inputs, one based on linear least squares, and one based on the recursive version of linear least squares. Through a series of experiments we show that while all three approaches significantly improve XCSF, least squares approaches appear to be best performing and most robust. Finally we show how XCSF can be extended to include polynomial approximations.