对撞机 | 是来搞数字医学人文的时候了

在最近这样一场没有硝烟的战争中，数据主要出现在三个叙事场景之中，一个是在有关实验室的基因分析中（基因数据库），一个是在疫情实时动态（感染死亡人数），另一个是对事件的回顾和物资的调配核算中（口罩分配）。实验室医学数据以其专业度确认了其适用范围和存在意义，疫情实时数据则以其权威性确认了其有效范围和存在价值，社会数据则以其舆论性确认了其传播范围和存在感。然而，正如WHO所提醒我们的，要记住他们是人，而不是数字。每一个数据的产生都有着人性的判断和人为的因素。这些决定了数据的存在，但历史往往就是由那些没有细节的数据构成的，个体经历在大历史的书写中往往并没有更多的意义。尤其是这样一个数字时代，技术看似磨平了经验与知识的凹凸度，不同质的一切都成为可以在一个网络中一个维度中被讨论和分析，并且用一种尺度去衡量。那么人在哪里？

Kirsten Ostherr的回答是，现在是时候去考虑当下这个技术中介的医疗系统中承受苦难和被治愈的人的问题了。尤其考虑到这个医疗系统追求的是通过人工智能、机器学习和其他计算的方式去将人类交流和身体反应转化为细致的数据集。她给出的答案是“数字医学人文”（digital medical humanities）。医学人文已经不是一个新鲜概念，只不过在大的数字人文浪潮中，Kirsten通过加上digital而对之进行了改造。在《健康人文的研究方法》（Research Methods in Health Humanities）（牛津大学出版社，2019）的《数字健康人文》（Digital Health Humanities）一章中，她指出，数字健康人文学科提供了一个“社会技术”框架（Wajcman 2002），该框架突出了技术开发和使用的社会层面。社会技术框架有助于检验和干预这些技术，以助他们提供更公正、公平和人道形式的医疗保健。她认为尽管数字技术在基因编辑、电子医疗档案（electronic health records）以及医生在线问诊方面得到了广泛使用，但与此同时，“病人的声音”与叙事也被淹没在健康大数据之中。如何发现病人的声音，如何让他们不仅仅成为一个简单的数字，这或许是计算技术与人文相结合的数字人文所能做的事情。Kirsten认为像数据可视化、文本挖掘，数据映射及网络抓取等数字人文常用的方法与人文价值、伦理和批判性思考的结合，是可以促使人文学者与计算机科学家联手去开发出可以理解人文历史、经验和文化现象的方法与工具。这一点，在这几天的中国数据分析的帖子中实际上也已经体现，但如何将之沉淀、记录与保存起来，让之成为历史的一部分，成为未来的人们可以在回顾历史时可以信任和使用的数据，则是研究者所必须考虑的事情。不要让一切在发生中已然消失。

The digital transformation of industries like banking, travel, and entertainment is old news, leaving many to ponder why healthcare has not yet been similarly disrupted. While change has been slower to come to medicine, digital technologies have nonetheless already opened up new techniques from gene editing to open electronic health records (EHRs) to patient reviews of doctors on Yelp. Inside of clinical spaces, the practice of medicine is now heavily mediated by screens, and outside, the everyday pursuit of health is heavily mediated by devices of quantification. Under these circumstances, almost any behavior or exposure that can be sensed and digitally quantified becomes reframed as a health behavior available for datafication, intervention, and optimization. This emergent digital health ecosystem produces new concepts of health, disease, risk, privacy, surveillance, and care. It also raises questions about what becomes of the human dimensions of suffering and healing in techno-mediated medical systems that pursue scalable clinical augmentation through artificial intelligence, machine learning, and other computational approaches seeking to convert the nuances of human communication and bodily expression into crunchable datasets.   

These conditions produce a field ripe for both humanistic and digital intervention - digital medical humanities. Like the field of critical code studies, this emerging field integrates historical and theoretical frameworks with applied interventions that aim not only to critique, but also to transform their objects of inquiry.

Much work in the medical humanities is premised on the idea that patients’ voices must be better represented to help address structural inequalities that are elided in narrowly biomedical approaches to care. In the era of EHRs, this is both a human problem and a computational problem. While one doctor can listen to one patient at a time, AI can listen to thousands or millions of patients at a time. Though I am skeptical that current Natural Language Processing algorithms are capable of accurately interpreting the subtleties of human communication, I am convinced that significant patterns and revelations could be found if we were capable of “distant reading” the patient narratives buried in healthcare’s big data. It is undoubtedly true that big data analytics need to develop better human contextual sensitivity to produce meaningful results. It is also true that the capacity to interpret personal health data in relation to population-scale data can yield discoveries that benefit patients precisely because they situate the individual in relation to the mass.

It is worth noting that the field of health informatics already uses many of the techniques adopted by digital humanities, such as data visualization, text mining, data mapping, and web scraping. In the present moment, analysis of big health data needs digital humanities methods and especially, critical insights into the values, ethics, and harms that are often embedded in the seemingly neutral operations of binary code. Digital humanists have learned to collaborate with computer scientists to develop new methods for understanding human history, experience, and cultural production. Now, both of these groups should join forces with medical humanists to design techniques for better understanding human experiences of illness through big health data. Part of this project must entail critical examination and intervention into the design of the very devices that millions of Americans cheerfully wear to quantify our own health and wellbeing. As technologies at the forefront of digital health, wearables promise to deliver grand insights about how exposures and behaviors shape health outcomes. As technologies that also surreptitiously engage in digital profiling, wearables compromise our privacy and autonomy, exploiting us through gender-, racial- and income-based manipulations, ultimately threatening to reproduce harmful patterns of health disparities through the privatization of care. There is a clear opening here for critical insights and creative energy from scholars whose digital interventions can help make medicine more humane, socially just, and equitable. What's more, right now the tech companies are actually listening.