内容来源：https://www.sciencenews.org/article/ai-violent-crime-forensics-blowflies

内容总结：

美国科学家近日研发出一种新型法医鉴定技术，通过分析案发现场常见的绿头苍蝇蛹壳，可在90秒内精准锁定昆虫种类，从而为推断死者遇害时间、案发地点及死因提供关键线索。

路易斯安那州立大学有机化学家拉比·穆萨团队将人工智能与化学检测技术相结合，利用场解吸质谱法捕捉不同蝇种蛹壳独特的化学成分，再通过机器学习模型进行比对识别。该技术已实现对19种不同地域采集的未知蛹壳进行100%准确鉴定。

研究表明，蛹壳作为昆虫发育过程中的坚硬外壳，能在尸体腐烂后长期保存。若在案发现场发现非本地蝇种蛹壳，则提示尸体可能被转移。此外，蛹壳中残留的化学成分不仅能反映昆虫发育周期，还可能包含受害者生前摄入的毒素信息。

英国基尔大学分析化学家福尔科·德里杰夫特评价称，这项技术突破了传统DNA分析受环境降解限制的瓶颈，为陈年悬案的法医调查开辟了新途径。穆萨教授表示，蛹壳中的化学分子如同记录犯罪信息的语言，未来通过研究分子随时间衰变的规律，甚至能推断长达十五年以上的死亡时间。

中文翻译：

梅根·罗森 撰
案发现场的丽蝇线索或许能帮助揭示受害者的死亡时间与其他谋杀细节——甚至在案发多年后仍能发挥作用。这些昆虫在尸体上繁育时会产下虫卵，孵化成熟后留下具有指示意义的蜕壳。这种被称为蛹壳的残留物，能帮助调查人员根据昆虫发育至蛹壳阶段所需时间，反向推算死者遇害时间。

但不同种类的昆虫发育速率存在差异。巴吞鲁日路易斯安那州立大学的有机化学家拉比·穆萨指出，要准确推断死亡时间，关键在于确定涉案昆虫的具体种类。目前她的研究团队已开发出一种快速鉴定技术：通过结合人工智能工具与化学检测，研究人员能在90秒内通过蛹壳完成蝇种鉴定。该研究成果于10月1日发表在《法医化学》期刊。

未参与该研究的英国基尔大学分析化学家法尔科·德里杰富特表示，蛹壳不仅能确定死亡时间，还能提供其他破案线索，例如尸体是否被移动过。"蛹壳会始终伴随尸体，"他解释道，"若在案发现场发现栖息地遥远的蝇种蛹壳，则表明尸体曾被转移。"

丽蝇堪称死亡探测专家，能在几分钟内发现两公里内的尸体。这些昆虫以尸体为食源产卵，孵化出的蛆虫在成熟过程中会化蛹，形成穆萨所说的"坚固耐用的微型结构"——蛹壳，成蝇最终将破壳而出。

卵、幼虫、蛹及蛹壳都能提供死亡时间信息，前提是能准确鉴定物种。这在早期阶段并非易事，因不同物种此时往往形态相似。穆萨指出，传统鉴别方法需从尸体采集虫卵或幼虫培育至成虫阶段，但若仅存蛹壳则此法失效。DNA分析是另一途径，但若昆虫样本暴露于自然环境导致DNA降解，检测就会变得困难。

穆萨团队提出了更简捷的解决方案：利用场解吸质谱技术绘制每种昆虫特有的蛹壳化学指纹，再通过机器学习预测模型处理这些化学数据以实现物种识别。德里杰富特评价称，该质谱技术能捕获其他化学检测方法通常遗漏的化合物，为物种鉴定提供了更丰富的数据支撑。

研究团队利用实验室培育的数百个丽蝇蛹壳化学指纹训练模型后，对来自全国各地的19个未知蛹壳进行盲测。结果显示，人工智能模型每次都能准确识别所属丽蝇物种。

该化学检测技术还能捕捉可存留数年的化学物质。穆萨表示，若能绘制这些分子随时间变化的衰减曲线，未来或可开发出通过蛹壳推断多年前命案死亡时间的新技术，从而判断"遗骸在此存放约15年而非两年"。

穆萨团队检测的分子能携带包括死亡时间、地点乃至死因在内的犯罪线索——蛹壳可能含有受害者摄入的毒物。她比喻道："这些分子就像特殊语言，只要你懂得解读，就能提取出海量信息。"

英文来源：

By Meghan Rosen
Crime scene clues from blowflies may help reveal a victim’s time of death — and other murderous details — perhaps even years later.
When colonizing a dead body, these insects lay eggs that mature into adult flies, leaving behind telltale remnants. The remnants, called puparial casings, could help investigators back calculate when someone died, based in part on the time it takes for insects to reach the casing stage.
But different species mature at different rates. To accurately estimate time of death, figuring out which species you’re dealing with is crucial, says Rabi Musah, an organic chemist at Louisiana State University in Baton Rouge.
Now, her team has developed a rapid method to do just that. By combining an AI tool with chemical detection, researchers can identify fly species from their casings — all within about 90 seconds, Musah’s team reports October 1 in Forensic Chemistry.
Time of death isn’t the only thing these casings could help determine, says Falko Drijfhout, an analytical chemist at Keele University in England who was not involved with the work. They could also offer other clues about a crime, like whether a body has been moved. “Casings will remain with the corpse,” he says. If investigators find casings from a species that lives far away, that’s a sign the body has been relocated.
Blowflies are death-detecting wizards. They can spot a dead body within minutes and from up to two kilometers away. These insects lay eggs on the body, which serves as a food source for the maggots that later hatch. As the maggots mature, their bodies transform into pupae, which are housed by puparial casings — “sturdy, hardy little structures,” Musah says. The adult fly ultimately emerges from these casings.
Eggs, maggots, pupae and casings can all offer information about a death’s timing — if you can identify the species. That’s not always easy to do, because at these early stages, different species tend to look alike. One way to tell them apart is to capture eggs or maggots from the body and then rear them to adulthood, when it’s easier to identify individual species. But that doesn’t work if you’ve only got casings, Musah says. Another way to differentiate species is by DNA analysis. But that can be tricky if insect material has been exposed to the elements and DNA has degraded.
Musah’s team came up with a simpler solution. The researchers mapped insect casings’ chemical fingerprints, which are unique to each species, using a technique called field desorption-mass spectrometry. Then, the team used a machine learning prediction model, a type of artificial intelligence, to crunch that chemical data and churn out the species identity.
The researchers’ mass spectrometry technique let them tap into compounds not typically captured by other chemical detection methods, Drijfhout says. That gave the team more data to work with, which is helpful for species identification.
Musah and her colleagues had trained their model on chemical fingerprints of hundreds of casings, sourced from lab-raised blowflies. Then, the team tested the system on 19 other casings that it had never seen before and had been collected across the country. The AI model identified the blowfly species the casings belonged to correctly each time, Musah’s team reported.
The researchers also showed that their chemical detection technique could pick up chemicals that tend to stick around for years. If scientists could map how these molecules weather over time, that could one day offer a new tool for estimating time of death from puparial casings, long after a death has occurred. That would let you say, “these remains have likely been here 15 years as opposed to two years,” Musah says.
The molecules Musah’s team detects can carry clues about crimes, like time and location of death and perhaps even cause — casings can contain poisons ingested by victims. These molecules are like a language, she says. “If you’re listening, there’s all this information you can extract.”