Discovering design principles of collagen molecular stability using a genetic algorithm, deep learning, and experimental validation.

Collagen is the most abundant structural protein in humans, providing crucial mechanical properties, including high strength and toughness, in tissues. Collagen-based biomaterials are, therefore, used for tissue repair and regeneration. Utilizing collagen effectively during materials processing ex vivo and subsequent function in vivo requires stability over wide temperature ranges to avoid denaturation and loss of structure, measured as melting temperature (Tm). Although significant research has been conducted on understanding how collagen primary amino acid sequences correspond to Tm values, a robust framework to facilitate the design of collagen sequences with specific Tm remains a challenge. Here, we develop a general model using a genetic algorithm within a deep learning framework to design collagen sequences with specific Tm values. We report 1,000 de novo collagen sequences, and we show that we can efficiently use this model to generate collagen sequences and verify their Tm values using both experimental and computational methods. We find that the model accurately predicts Tm values within a few degrees centigrade. Further, using this model, we conduct a high-throughput study to identify the most frequently occurring collagen triplets that can be directly incorporated into collagen. We further discovered that the number of hydrogen bonds within collagen calculated with molecular dynamics (MD) is directly correlated to the experimental measurement of triple-helical quality. Ultimately, we see this work as a critical step to helping researchers develop collagen sequences with specific Tm values for intended materials manufacturing methods and biomedical applications, realizing a mechanistic materials by design paradigm.

Transforming text to music using artificial intelligence improves the frontal lobe function of normal older adults.

INTRODUCTION: Recent advances in artificial intelligence (AI) have been substantial. We investigated the effectiveness of an online meeting in which normal older adults (otokai) used a music-generative AI that transforms text to music (Music Trinity Generative Algorithm-Human Refined [MusicTGA-HR]). METHODS: One hundred eighteen community-dwelling, cognitively normal older adults were recruited through the internet (64 men, 54 women; mean age: 69.4 ± 4.4 years). Using MusicTGA-HR, the participants chose music that they thought was the most suitable to a given theme. We established 11 classes of 7-10 members and one instructor each. Each class held an online meeting once a week, and each participant presented the music they chose. The other participants and the instructor then commented on the music. Neuropsychological assessments were performed before and after the intervention for 6 months, and the results before and after the intervention were statistically analyzed. RESULTS: The category and letter word fluencies (WFs) were significantly improved (category WF: p = .003; letter WF: p = .036), and the time of the Trail-Making Test-B was also significantly shortened (p = .039). The Brain Assessment, an online cognitive test we developed, showed significant improvement in the memory of numbers (p < .001). CONCLUSION: The online meeting of the otokai, which used music-generative AI, improved the frontal lobe function and memory of independent normal older adults.

Quantum conditional generative adversarial network based on patch method for abnormal electrocardiogram generation.

To address the scarcity and class imbalance of abnormal electrocardiogram (ECG) databases, which are crucial in AI-driven diagnostic tools for potential cardiovascular disease detection, this study proposes a novel quantum conditional generative adversarial algorithm (QCGAN-ECG) for generating abnormal ECG signals. The QCGAN-ECG constructs a quantum generator based on patch method. In this method, each sub-generator generates distinct features of abnormal heartbeats in different segments. This patch-based generative algorithm conserves quantum resources and makes QCGAN-ECG practical for near-term quantum devices. Additionally, QCGAN-ECG introduces quantum registers as control conditions. It encodes information about the types and probability distributions of abnormal heartbeats into quantum registers, rendering the entire generative process controllable. Simulation experiments on Pennylane demonstrated that the QCGAN-ECG could generate completely abnormal heartbeats with an average accuracy of 88.8%. Moreover, the QCGAN-ECG can accurately fit the probability distribution of various abnormal ECG data. In the anti-noise experiments, the QCGAN-ECG showcased outstanding robustness across various levels of quantum noise interference. These results demonstrate the effectiveness and potential applicability of the QCGAN-ECG for generating abnormal ECG signals, which will further promote the development of AI-driven cardiac disease diagnosis systems. The source code is available at github.com/VanSWK/QCGAN_ECG.

A two-stage Bayesian adaptive design for minimum effective dose (MinED)-based dosing-finding trials.

Conventional phase I designs for finding a phase II recommended dose (P2RD) based on toxicity alone is problematic because the maximum tolerated dose (MTD) is not necessarily the optimal dose. Instead, recently attention has been given to find the minimum effective dose (MinED) - defined as the lowest effective dose. Traditional paradigms for the MinED studies are conducted as dose-ranging or dose-response trials which involve several doses and randomize patients among doses to find the MinED. An alternative approach for the MinED study is the so-called MinED-based dose-finding study, in which instead of conducting hypothesis testings and without power analysis, this kind of trial conduct dose escalation/de-escalation to target a pre-set MinED target. In this study, we propose a new Bayesian two-stage adaptive design schema based on framework of the interval-based phase I method. The proposed method is model-free without curve pre-specifications, which is suitable for various dose-efficacy relationships. The proposed method shows desirable theoretical finite property of semi-coherence and large sample property of consistency. A random scenario generative algorithm for the MinED has also been proposed for extensive simulation studies, which demonstrated desirable performances of the proposed method. An R package "MinEDfind" and a Shiny app have been developed for implementing the method.