{"id":14808,"date":"2023-06-01T10:01:13","date_gmt":"2023-06-01T10:01:13","guid":{"rendered":"https:\/\/gustavopineda.com\/?p=14808"},"modified":"2023-06-01T10:01:13","modified_gmt":"2023-06-01T10:01:13","slug":"ai-healthcare-integration","status":"publish","type":"post","link":"https:\/\/gustavopineda.com\/?p=14808","title":{"rendered":"This children’s hospital is integrating AI with healthcare"},"content":{"rendered":"
<\/p>\n

Much as hospitals became major vectors for the spread of COVID<\/a> in early 2020, hospitals are already starting to host another viral phenomenon still in its infancy: generative AI in the workplace<\/a>. Highly-ranked healthcare facilities like Boston Children\u2019s Hospital, connected as they are to major research institutions, are some of the most prominent customer-facing operations in healthcare industry. <\/p>\n

And given that healthcare represents about 18 percent of the U.S. GDP<\/a>, of course <\/em>these organizations will want to take advantage of the latest technology that promises a revolution in productivity.\u00a0<\/p>\n

Boston Children\u2019s Hospital, consistently ranked among the best children\u2019s hospitals in the U.S.,<\/a> employs a \u201cChief Innovation Officer,\u201d John Brownstein, an epidemiologist who runs a division called the Innovation & Digital Health Accelerator<\/a>. Brownstein\u2019s past work combining technology and health includes the creation of a site called \u201cFlu Near You,\u201d which was repurposed during the early days of the pandemic as \u201cCovid Near You<\/a>\u201d for obvious reasons, according to New York Times Magazine<\/em><\/a>. It still exists in a more general form as \u201cOutbreaks Near Me<\/a>.\u201d It\u2019s an unsettlingly useful website for tracking pathogens. \u00a0<\/p>\n

And now Brownstein is turning his attention to AI<\/a>.<\/p>\n

First things first, according to Brownstein: from his standpoint there\u2019s no need to lay anyone off just because AI is invading healthcare. \u201cThis is not meant as a replacement for the human,\u201d Brownstein told Mashable in an April interview. \u201cThis is an augmentation. So there’s always a human in the loop.\u201d\u00a0<\/p>\n

\n SEE ALSO:<\/span>
\n
\n What not to share with ChatGPT if you use it for work
\n <\/use><\/svg>
\n <\/a>\n <\/div>\n

In April, as prompt engineering<\/a> became a buzzworthy new tech job, Boston Children\u2019s tipped its hand to the public about the fact that change was afoot when it posted a job ad<\/a> seeking a prompt engineer of its own. In other words, the hospital was hiring a specialist to train AI language models that can improve hospital operations, and in theory, this person is supposed to improve conditions for hospital staff.<\/p>\n

According to Brownstein, that\u2019s because his department has a directive to reduce \u201cprovider burnout.\u201d Boston Children\u2019s has what he called \u201can internal team that builds tech.\u201d Their job, he explained, is to locate places in \u201cthe world of work\u201d where technology can play a role, but isn\u2019t yet. They literally sit in \u201cpain points\u201d within Boston Children\u2019s Hospital, and devise ways to, well, ease the pain.<\/p>\n

What this means in practice is a bit mind-bending.<\/p>\n

Easing the pain with AI\u00a0<\/h2>\n

One \u201cpain point\u201d in any hospital is directing patients from point A to point B, a tough exercise in communication that can include speed bumps like confusion due to illness or stress, or language barriers. \u201cAlready out of the gate, we can query ChatGPT with questions about how to navigate our hospital,\u201d Brownstein said. \u201cIt’s actually shocking, what these are producing without any amount of training from us.\u201d\u00a0 ChatGPT \u2014 and not some future version but the one you already have access to \u2014 can tell you how to get around \u201cnot just our hospital, but any hospital,\u201d according to Brownstein.<\/p>\n

So it\u2019s more than realistic to imagine a machine kiosk that patients can receive useful answers to questions like, Brownstein offered, \u201cWhere can I pray?\u201d And it\u2019s probably also the hope of many healthcare workers that they don\u2019t have to be stopped in their tracks with questions like that. Not everyone is a people person.<\/p>\n

But Brownstein also has ideas for new ways providers can use patient data thanks to AI.<\/p>\n

The idea that AI will be involved in the processing of actual patient data set off alarms for Mildred Cho professor of pediatrics at Stanford\u2019s Center for Biomedical Ethics. After reviewing the prompt engineer job ad, she told Mashable, \u201cWhat strikes me about it is that the qualifications are focused on computer science and coding expertise and only \u2018knowledge of healthcare research methodologies\u2019 while the tasks include evaluating the performance of AI prompts.\u201d<\/p>\n

\u201cTo truly understand whether the outputs of large language models are valid to the high standards necessary for health care, an evaluator would need to have a much more nuanced and sophisticated knowledge base of medicine and also working knowledge of health care delivery systems and the limitations of their data,\u201d Cho said.\u00a0<\/p>\n

\n SEE ALSO:<\/span>
\n
\n ChatGPT-created resumes are dealbreakers for recruiters
\n <\/use><\/svg>
\n <\/a>\n <\/div>\n

Cho further described a nightmare scenario: what if the prompt engineer helps retrain a language model, or tweak an automated process, but due to faulty assumptions.\u00a0For instance, what if they train racial bias, or other persistent mistakes into it. Given that all data collected by people is inherently flawed, a shiny new process could be built on a foundation of errors.<\/p>\n

\u201cOur prompt engineer is not going to be working in a bubble,\u201d Brownstein said. His team devotes time, he said, to worrying about \u201cwhat it means to have imperfect data.\u201d He was confident that the process wouldn\u2019t be \u201cput a bunch of data in and like hope for the best.\u201d<\/p>\n

Using AI to customize discharge instructions<\/h2>\n

But lest we forget, \u201cput in a bunch of data and hope for the best\u201d is an apt description of how large language models work, and the results are often, well, awful<\/a>.\u00a0<\/p>\n

For an example where the data need to be right-on-the-money, look no further than Brownstein\u2019s absolutely fascinating vision for the discharge instructions of the future. You\u2019ve probably received \u2014 and promptly thrown away \u2014 many discharge instructions. <\/p>\n

Perhaps you got a bump on the head in a car accident. After getting checked out at the hospital and being cleared to go home, you likely received a few stapled pages of information about the signs of a concussion, how to use a cold compress, and how much ibuprofen to take.\u00a0<\/p>\n

With an LLM trained on your individual patient information, Browstein said, the system knows, among other things where you live, so it can tell where to go to buy your ibuprofen, or not to buy Ibuprofen at all, because you\u2019re allergic. But that\u2019s just the tip of the iceberg.\u00a0<\/p>\n

\u201cYou’re doing rehab, and you need to take a walk. It’s telling you to do this<\/em> walk around this particular area around your house<\/em>. Or it could be contextually valuable, and it can modify based on your age and, and various attributes about you. And it can give that output in the voice, that is the most compelling to make sure that you adhere to those instructions.\u201d<\/p>\n

New tech historically has found its way into hospitals quickly\u00a0<\/h2>\n

David Himmelstein, a professor in the CUNY School of Public Health and a prominent critic of the U.S. for-profit healthcare system said that while he had heard about potential uses of AI in hospitals that concerned him, this one didn\u2019t strike him as \u201coffensive.\u201d He noted that discharge instructions are \u201calmost boilerplate\u201d anyway, and seemed unconcerned about the potential change.<\/p>\n

However, he worries about what such systems could mean for privacy. \u201cWho gets this information?\u201d he wondered. \u201cSounds like it puts the information in the hands of Microsoft \u2014 or Google if they use their AI engine.\u201d\u00a0<\/p>\n

In widespread use, these are major concerns for hospitals moving forward, but Brownstein said that Boston Children\u2019s Hospital, for its part, \u201cis actually building internal LLMs,\u201d meaning it won\u2019t rely on companies like Google, Microsoft, or ChatGPT parent company OpenAI. \u201cWe actually have an environment we’re building, so that we don’t have to push patient data anywhere outside the walls of the hospital.\u201d\u00a0<\/p>\n

Himmelstein further pointed out that systems for automating hospitals are far from new, and\u00a0 have not created places bureaucracy-free paradises, where work runs smoothly and efficiently, even though Himmelstein noted, companies like have been making such promises since the 1960s. He provided a fascinating historical document to illustrate this point: an IBM video from 1961 that promises electronic systems that will slash bureaucracy and \u201celiminate errors.\u201d<\/p>\n

\n
<\/div>\n<\/div>\n

But in the month since Mashable first spoke to Brownstein, the AI situation has progressed at Boston Children\u2019s Hospital. In an email, Browstein reported \u201ca ton of progress\u201d on large language models, and \u201cincredible\u201d prompt engineer in the process of being onboarded.\u00a0 \u00a0<\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"

Much as hospitals became major vectors for the spread of COVID in early 2020, hospitals are already starting to host another viral phenomenon still in its infancy: generative AI in…<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"registered_only","ping_status":"closed","sticky":false,"template":"","format":"gallery","meta":{"nf_dc_page":"","om_disable_all_campaigns":false,"_monsterinsights_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[17],"tags":[],"class_list":["post-14808","post","type-post","status-publish","format-gallery","hentry","category-technology","post_format-post-format-gallery"],"aioseo_notices":[],"jetpack_featured_media_url":"","jetpack-related-posts":[],"jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/gustavopineda.com\/index.php?rest_route=\/wp\/v2\/posts\/14808","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/gustavopineda.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/gustavopineda.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/gustavopineda.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/gustavopineda.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=14808"}],"version-history":[{"count":0,"href":"https:\/\/gustavopineda.com\/index.php?rest_route=\/wp\/v2\/posts\/14808\/revisions"}],"wp:attachment":[{"href":"https:\/\/gustavopineda.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=14808"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/gustavopineda.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=14808"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/gustavopineda.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=14808"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}