/static/upload/image/20240511/1715409640121861.png

Research 研究

AI for Science & AI for Medicine

实验室研究方向:

1)多模态生物模型

2)神经元网络的结构和动力学

3)计算机视觉技术在生物医学,尤其是神经退行性病中的应用


生物医学领域正经历着前所未有的深刻范式变革,这核心推动力量正是大模型技术的崛起和应用。在这个充满机遇与挑战的时代,本实验室将超分辨成像和计算机视觉模型作为研究的切入点,深入探索多尺度跨模态神经元网络的结构和动态,分析神经元网络的结构动力学机制和及其在疾病模型中的作用,以期在生物医学研究中取得突破性的进展。

我们的研究方向聚焦于多尺度跨模态神经元网络,这是一个复杂而精细的生物系统,其结构和动态对于理解大脑功能和神经性疾病的发病机制至关重要。通过超分辨成像技术,我们能够以前所未有的精度和分辨率观察到神经元网络的微观结构,捕捉到生物分子之间的相互作用和动态变化。而计算机视觉模型则为我们提供了一种高效、准确的数据分析和处理手段,能够从海量的图像数据中提取出有价值的信息,帮助我们揭示神经元网络的复杂性和规律性。

在这个基础上,我们将开发大模型来推动“AI for Science”和“AI for Medicine”的研究。大模型凭借其强大的学习和推理能力,能够处理和分析海量的生物医学数据,发现隐藏在数据中的规律和模式,为生物医学研究提供新的思路和方法。通过大模型的应用,我们可以更深入地理解神经元网络的结构和功能,预测神经性疾病的发病机制,为疾病的早期诊断和治疗提供科学依据。

然而,我们也清楚地认识到,这一研究面临着诸多挑战。首先,生物医学数据多尺度、多模特的复杂性和多样性要求我们在数据获取和处理方面不断创新和改进。其次,超分辨成像技术和计算机视觉模型的结合需要我们深入探索算法设计和优化,以提高图像分析和信息提取的准确性和效率。此外,大模型的训练和应用需要强大的计算资源支持,我们需要不断探索新的计算技术和架构,以满足研究的需要。

展望未来,我们将继续深化对多尺度跨模态神经元网络的研究,推动AI在生物医学领域的应用和发展。我们期待通过跨学科(数学、物理、工程、生物学和医学)的合作和技术的创新,解决当前面临的挑战,为生物医学研究的深入发展贡献我们的力量。同时,我们也相信,随着大模型技术的不断进步和应用,生物医学领域将迎来更加美好的未来。

2024-05-11 979

Research Focus:

  1. Multimodal biological models

  2. Structure and dynamics of neuronal networks

  3. Applications of computer vision in biomedicine, particularly in neurodegenerative diseases

Biomedical research is undergoing a profound paradigm shift driven by the rise and application of AI, particularly the recent development of large language models (LLMs). We aim to develop large models to advance "AI for Science" and "AI for Medicine." Our laboratory uses multi-modal data, including super-resolution imaging, electrophysiology, omics, and LLMs, as entry points to explore the structure, dynamics, and functions of multi-scale neuronal networks. This approach focuses on analysing the structural dynamics mechanisms of neuronal networks and their roles in disease models.


These models, with their powerful learning and reasoning capabilities, can process and analyse massive biomedical data, potentially uncovering hidden patterns and providing new insights. This approach will deepen our understanding of neuronal network structures and functions, predict mechanisms of neurological diseases, and offer scientific bases for early diagnosis and treatment.


This is certainly challenging work, given the complexity and diversity of multi-scale, multimodal biomedical data, which requires continuous innovation in data acquisition and processing. Combining imaging data with sequential data necessitates advanced algorithm design and optimisation. Additionally, training and applying LLMs demand substantial computational resources, prompting the exploration of new computing technologies and architectures.

Through interdisciplinary integration (mathematics, physics, engineering, biology, AI and medicine) and technological innovation, we aim to address current challenges and contribute to the advancement of biomedical research. We welcome students and scholars from the above backgrounds to visit and collaborate!

 



扫描二维码

手机扫一扫添加微信