标记可能缓解大语言模型 回答医疗问题的过度自信:定量研究
Raphaël Bentegeac1,2, Bastien Le Guellec3,4, Grégory Kuchcinski3,4
1Department of Public Health, Lille University, Lille University Hospital Center, avenue du Professeur Emile Laine, Lille, 59037, France.
Journal of medical Internet research
|August 29, 2025
概括
医疗聊天机器人通常表达出高度的信心, 而不是自我评价的确定性, 能够更好地预测聊天机器人的医疗问题准确性,
科学领域:
- 医学的人工智能
- 自然语言处理
- 医疗信息学
背景情况:
- 聊天机器人在医学上表现有前途,
- 然而,由于对错误答案的过度自信,
研究的目的:
- 将代币概率与聊天机器人预测医疗反应准确性的表达信心进行比较.
- 在评估聊天机器人的性能时,评估代币概率是否优于自我报告的信心.
主要方法:
- 九个大型语言模型 (LLM) 回答了2522个美国医疗执照考试问题.
- 记录和分析了模型的信心和响应令牌概率.
- 使用AUROC,校准误差和Brier分数来评估预测性能.
主要成果:
- 聊天机器人的准确性各不相同,GPT-4o达到89%,Phi-3-Mini达到56.5%.
- 表达的信心预测准确度很差 (AUROC 0.52 - 0.68).
- 代币概率始终超过了信心 (AUROC 0.71-0.87),表明更好的准确性预测.
结论:
- 聊天机器人在医疗环境中难以准确的自我评估.
- 提供更可靠的方法来评估聊天机器人响应的准确性.
- 临床医生不应该依赖聊天机器人自我评价的确定性; 象征性概率是一个更好的选择.
更多相关视频
相关概念视频
Improving Translational Accuracy
11.8K
Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...
11.8K
The Availability Heuristic
6.4K
A heuristic is a general problem-solving framework (Tversky & Kahneman, 1974). You can think of these as mental shortcuts that are used to solve problems. Different types of heuristics are used in different types of situations, and the impulse to use a heuristic occurs when one of five conditions is met (Pratkanis, 1989):
6.4K
Sensitivity, Specificity, and Predicted Value
658
In healthcare diagnostics, laboratory tests play a crucial role in identifying and diagnosing a wide range of medical conditions. However, interpreting test results is not always straightforward. An abnormal test result does not always confirm the presence of a disease, just as a normal result does not guarantee its absence. To assess the reliability of these diagnostic tools, healthcare practitioners rely on two key statistical indicators: sensitivity and specificity.
Sensitivity is the...
Sensitivity is the...
658
Uncertainty: Confidence Intervals
4.6K
The confidence interval is the range of values around the mean that contains the true mean. It is expressed as a probability percentage. The interpretation of a 95% confidence interval, for instance, is that the statistician is 95% confident that the true mean falls within the interval. The upper and lower limits of this range are known as confidence limits. The confidence limits for the true mean are estimated from the sample's mean, the standard deviation, and the statistical factor...
4.6K
Microorganisms in Medicine and Therapeutics
312
Microorganisms play a fundamental role in vaccine development, gene therapy, and therapeutic production. Their biological properties are harnessed to advance medicine and public health. Beyond immunization, microorganisms contribute to gut health, antibiotic synthesis, and genetic disease treatment.Live Attenuated and Inactivated VaccinesLive attenuated vaccines, such as the measles, mumps, and rubella (MMR) vaccine, utilize weakened forms of pathogens to closely resemble natural infections.
312
Testing a Claim about Population Proportion
3.4K
A complete procedure for testing a claim about a population proportion is provided here.
There are two methods of testing a claim about a population proportion: (1) Using the sample proportion from the data where a binomial distribution is approximated to the normal distribution and (2) Using the binomial probabilities calculated from the data.
The first method uses normal distribution as an approximation to the binomial distribution. The requirements are as follows: sample size is large...
There are two methods of testing a claim about a population proportion: (1) Using the sample proportion from the data where a binomial distribution is approximated to the normal distribution and (2) Using the binomial probabilities calculated from the data.
The first method uses normal distribution as an approximation to the binomial distribution. The requirements are as follows: sample size is large...
3.4K


