一种人工鼻子或许能一早嗅出帕金森病

An artificial nose may be able to detect Parkinson’s disease early

一种人工鼻子或许能一早嗅出帕金森病

The inspiration came from the strange case of a retired nurse

灵感来自一位退休护士的奇特本领【新知】

PARKINSON’S DISEASE is a progressive neurological disorder caused by the deaths of neurons in parts of the brain called the substantia nigra. Symptoms include loss of motor control, mood disturbance, ruined sleep and altered sense of smell. It is incurable. Early medical intervention can, however, relieve these symptoms and prolong survival. That makes prompt diagnosis desirable.

帕金森病是一种因脑内黑质区域中神经元的死亡引起的进行性神经失调。其症状包括运动控制丧失、情绪紊乱、睡眠障碍和嗅觉改变。它无法治愈。但是尽早施加医疗干预可以缓解症状，延长生存期。这就使得及早确诊很有益处。

Unfortunately, the initial signs of Parkinson’s vary from person to person and there is no specific test at this early stage which can reliably distinguish it from other brain illnesses. It therefore often goes undetected until clear and characteristic manifestations, such as tremors and slowed body movement, appear. But that may soon change. A study published in ACS Omega, by Chen Xing and Liu Jun at Zhejiang University, in China, describes an invention which may be able to detect Parkinson’s before the onset of tremulousness. The device in question is an artificially intelligent electronic nose.

不幸的是，帕金森病的最初症状因人而异，在发病的早期阶段，还没有哪种专门检测能可靠地将它和其他脑部疾病区分开来。因此它经常要等到出现明显且典型的临床表现，如颤抖和身体动作迟缓，才会被确诊。但这种情况可能很快就会改变了。浙江大学的陈星和刘军发表在《ACS Omega》期刊上的研究介绍了一项发明，或许能在震颤的症状出现前就检测出帕金森病。这款设备是一个人工智能电子鼻。

A whiff of trouble

一丝麻烦的气息

The quest the two researchers embarked on to build this nose began in 2019. That was when they heard reports of Joy Milne, a retired nurse living in Scotland, who could detect people with Parkinson’s from a distinctive odour they emitted—distinctive to her, at least, though for reasons still obscure, undetectable by others.

两位研究人员制造这个电子鼻的探索始于2019年。当时他们听到有报道称，住在苏格兰的退休护士乔伊·米尔恩（Joy Milne）能闻到帕金森患者散发出的一种特殊气味，从而发现患病者。这种气味至少对她来说是特殊的，不过其他人无法察觉，原因不明。

Mrs Milne first noticed this odour when her husband developed the illness. She made the general connection later, when she smelled it at sufferers’ support groups attended by her spouse. Tests using clothes worn by patients confirmed her ability. She even noted one seemingly healthy individual as having the disease months before other symptoms developed.

米尔恩第一次注意到这种气味是在她丈夫患病之后。后来，当她在丈夫参加的患者支持小组里闻到这种气味时，才意识到普遍的关联。用病人穿过的衣服进行的测试证实了她的能力。她甚至指出一个看似健康的人已经得病了，而此人要在几个月后才出现其他症状。

Carting Mrs Milne around the world to sniff patients who may have Parkinson’s is, however, not a practical option, so researchers working with her looked for the odour’s source, with a view to detecting it in some other way. They found it in sebum, an oily secretion produced by the skin. The sebum of those with Parkinson’s, they discovered, has unusually high concentrations of certain volatile organic compounds, including dodecane, acetone and ethyl acetate. When these are acted on by yeast cells which live naturally on the skin, the result is the mysterious odour.

不过，让米尔恩满世界去嗅出帕金森患者并不可行，所以研究人员和她一同寻找这种气味的来源，以期之后用其他方法来探测到它。他们在皮肤产生的一种油性分泌物皮脂中发现了这种气味。他们发现，帕金森病患者的皮脂中某些挥发性有机化合物的浓度异常地高，包括十二烷、丙酮和乙酸乙酯。当皮肤上自然存在的酵母细胞对它们起作用时，就会产生这种神秘的气味。

This Parkinson’s-specific list can be detected using a laboratory technique called gas chromatography-mass spectrometry (GCMS), which is promising—for GCMS machines are a bit more deployable than Mrs Milne is. Unfortunately, they are still too complex and expensive to be used routinely in clinics. However, Dr Chen and Dr Liu, who are both biomedical engineers, reckoned they could come up with a cheaper and more portable alternative.

这个帕金森病专有的物质清单可以用一种叫做气相色谱质谱联用（GCMS）的实验室技术检测出来，这就很有希望了，因为部署这种机器总比让米尔恩到处跑容易一些。遗憾的是，它们仍然过于复杂和昂贵，不能用作诊所的常规设备。不过，生物医学工程师陈星和刘军认为他们可以构造出更便宜也更便携的替代仪器。

The upshot is a machine not much bigger than a toaster. It turns a sample of sebum into a vapour that is then probed with sound waves. Different molecular mixtures have different effects on the waves, and, with a bit of training, a computer loaded with an appropriate piece of artificial intelligence can learn to recognise the pattern associated with sebum from Parkinson’s patients.

他们的成果是一台比烤面包机大不了多少的机器。它将一份皮脂样本转化为蒸汽，然后用声波探测。不同的分子混合物对声波有不同的影响，在稍加训练后，一台配备有恰当的人工智能的电脑就能学会识别与帕金森患者的皮脂相关联的模式。

The researchers tested their system on samples from 43 people with Parkinson’s disease and 44 who were healthy. They found it could correctly identify a Parkinson’s patient as having the disease about 70% of the time and a healthy control as being clear of it about 80%. That is not yet as good as Mrs Milne’s nose, which has a history of correctly identifying the presence of Parkinson’s all the time. But it is a start. If its reliability can be improved, the compactness and potential cheapness of the system Dr Chen and Dr Liu have come up with may eventually prove a boon for the early diagnosis of the illness. ■

研究人员对43名帕金森患者和44名健康受试者的样本进行了测试。他们发现，在病患组，70%的情况下该系统可以正确识别出患者，而在健康对照组，80%的情况下它可以排除掉未患病的受试者。这比不上米尔恩的鼻子——米尔恩一直以来识别帕金森患者的正确率是百分之百。但这只是一个开始。如果能提升可信度，那么陈星和刘军发明的这个小巧简洁、可能还会很便宜的系统或许最终会成为帕金森病早期诊断的一大福音。