对于不规则的多变量时间序列的可解释时间推理. 一个案例研究,用于早期预测多重药物耐药性
IEEE transactions on bio-medical engineering
|July 23, 2025
概括
我们开发了可解释的AI方法,从复杂的健康数据中预测患者的结果. 我们的方法提高了多药性耐药性 (MDR) 和循环失效的预测,为更好的患者护理提供了临床见解.
科学领域:
- 人工智能的人工智能
- 医疗保健信息学 医疗保健信息学
- 临床预测模型临床预测模型
背景情况:
- 多变量时间序列 (MTS) 数据在医疗保健中存在挑战,原因是不规则性和时间依赖性.
- 现有的"MTS-TS"模型往往缺乏临床解释性,阻碍了采用.
研究的目的:
- 为"MTS-TS"架构引入新的可解释的人工智能 (XAI) 方法.
- 为了能够跟踪患者的进化,并确定不良结果的关键变量.
- 评估关于多药性耐药性 (MDR) 和循环衰竭预测的框架.
主要方法:
- 开发了不规则的时间SHapley添加式扩展 (IT-SHAP) 用于后期分析.
- 实现了Hadamard注意力,用于内在时间依赖性捕获.
- 使用因果条件相互信息来预先选择特征.
主要成果:
- 使用哈达马德注意的GRU实现了MDR预测的高性能 (ROC-AUC=0.783).
- 在循环衰竭预测方面,LSTM表现优越 (ROC-AUC=0.9970).
- IT-SHAP确定了早期抗生素使用和细菌培养为关键风险因素,并得到临床医生的验证.
结论:
- 拟议的框架在"MTS-TS"模型中提供了时间可解释性.
- 临床医生可以追踪疾病的发展轨迹,并了解每个时间阶段的可变贡献.
- 整合到电子健康记录系统可以改善早期干预,抗菌药物管理和感染控制.
更多相关视频
04:09Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma
Published on: October 10, 2018
8.3K
12:18A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
7.6K
相关概念视频
Drug Concentration Versus Time Correlation
1.2K
The plasma drug concentration-time curve is a crucial tool in pharmacokinetics, representing the drug's concentration in plasma at different time intervals post-administration. This curve illustrates the drug's journey from absorption into the systemic circulation, distribution to body tissues, and eventual elimination through excretion or biotransformation.
Two pivotal parameters are the minimum effective concentration (MEC) and the minimum toxic concentration (MTC). The MEC is the...
Two pivotal parameters are the minimum effective concentration (MEC) and the minimum toxic concentration (MTC). The MEC is the...
1.2K
Time Course of Drug Effect
2.2K
The progression of a drug's impact can be analyzed by examining both the concentration-time course and the effect-time course. The concentration-time course is determined by the drug's half-life and is influenced by factors such as its pharmacokinetics, including absorption, distribution, metabolism, and elimination. The effect of the drug is often related to its concentration in the plasma and is calculated using the maximum drug effect and the plasma concentration that generates 50...
2.2K
Chronopharmacokinetics: Circadian Rhythms and Influence on Drug Response
130
Circadian rhythms are cyclic changes that are crucial in plasma drug concentrations. Various standard circadian parameters, including core body temperature, heart rate, and other cardiovascular factors, directly impact disease states and the therapeutic response to drug therapy.
The time of drug administration is an important factor to consider, as it can influence the toxic dose of a drug. For example, a study conducted by Prins et al. in 1997 examined the effects of the timing of...
The time of drug administration is an important factor to consider, as it can influence the toxic dose of a drug. For example, a study conducted by Prins et al. in 1997 examined the effects of the timing of...
130
Model Approaches for Pharmacokinetic Data: Distributed Parameter Models
128
Pharmacokinetic models are mathematical constructs that represent and predict the time course of drug concentrations in the body, providing meaningful pharmacokinetic parameters. These models are categorized into compartment, physiological, and distributed parameter models.
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
128
Steps in Outbreak Investigation
209
In the ever-evolving field of public health, statistical analysis serves as a cornerstone for understanding and managing disease outbreaks. By leveraging various statistical tools, health professionals can predict potential outbreaks, analyze ongoing situations, and devise effective responses to mitigate impact. For that to happen, there are a few possible stages of the analysis:
209
Treatment Resistant Cancers
3.4K
Cancer is the second leading cause of death in the United States. A cancer cell is genetically unstable and hence can mutate faster. They can also modify their microenvironment and escape immune surveillance. The difficulties in treating cancer are further compounded by the emergence of rapid resistance to anticancer drugs. The most common ways to attain resistance in cancer cells include alteration in drug transport and metabolism, modification of drug target, elevated DNA damage response, or...
3.4K
