Intensive care units around the world use daily diaries to track patients’ emotional as well as physical ups and downs. Many find that keeping these helpful resources can add to nurses’ already-heavy workloads. How might generative AI help? 

Researchers at University Hospital of Schleswig-Holstein in Germany count the ways.

“Several factors speak in favor of an AI-generated ICU diary, such as time savings, the standardized development of entries and the retrospective reconstruction of a diary from medical and nursing records,” write senior author Peter Nydahl, PhD, RN, and colleagues. “AI-generated ICU diaries could be beneficial to patients and their families, helping them cope with a challenging time in their lives,” they add. 

The paper is slated to run in the February 2025 edition of Intensive and Critical Care Nursing. Among the benefits the authors expound upon are these seven: 

1. AI-generated diaries save nursing time. 

An AI-generated diary provides a “very impressive overview of the patient’s stay” in the ICU and elsewhere in the hospital, Nydahl and co-authors write. They add that the resource can be created in minutes and given to every patient at discharge. 

‘An AI-generated diary supplies a compact summary of the patient’s treatment and all its key events.’ 

2. AI-generated diaries protect nurses’ privacy.

ICU nurses become involved in the privacy of their diarizing patients because they, the nurses, read and add entries themselves. “For some nurses, this can be an additional emotional burden, making it difficult to maintain the professional relationship,” the authors write before noting that AI-generated diaries can help overcome this barrier.

‘It should be important to note that the implementation of an AI-generated ICU diary is followed by the legal framework of the country. This data, like all hospital information, is highly valuable and should be treated with special protection.’

3. AI-generated diaries standardize diaries.

Another barrier against ICU diaries is the heterogenous use and delivery of diaries [6]. Currently, there are no specific recommendations for the length of the entries or the content or how the diary should be presented, which makes the usage of ICU diaries very heterogenous [14]. AI-generated ICU diaries can enable a standardized intervention. 

‘Standardizing the scope, presentation and conversion of AI reports will make them more comparable and protect them from the potential bias of personal influence by nurses.’

4. AI-generated diaries are impartial and objective. 

Personal perspectives and values may influence the nursing perspective on patients, their situation and their relatives, even on a subconscious level, the authors remind. By comparison, AI-generated diaries “can be impartial and as objective as possible.”

‘AI-generated ICU diaries can largely reduce the personal-perspective factor, although even AI is not free of bias.’

5. AI-generated diaries ensure continuity. 

In times of nursing shortage, staff often rotate between different ICUs, and many different persons and professions are involved in patients’ care, Nydahl and colleagues write. 

‘AI-generated diaries ensure continuity. The diary will have a homogeneous style, no breaks, likely no missing days and completely cover the critical-care journey.’ 

6. AI-generated diaries enable diaries-on-demand. 

Currently, most ICU diaries are written prospectively, the authors explain. “As a result, there is no way to get an ICU diary for the patients if the unit does not perform the intervention.”

‘AI can create an ICU diary on demand, without much effort for the nurses, physicians and therapists.’

7. AI-generated diaries save personal resources. 

An often-reported barrier against the use of ICU diaries is the demand they can place on already-overworked nurses. “AI-generated diaries spare these personal resources, support a reflective perspective in care, and ease stressful situations such as rushing an entry at the end of a shift.” 

‘Nurses write a short prompt, and everything else will be expanded with excellence by AI.’

The paper is available in full for free. 

Nabla Joins the Coalition for Health AI (CHAI) to Advance AI Governance in Healthcare - Nabla is joining forces with the Coalition for Health AI (CHAI), a diverse consortium of more than 3,000 organizations, including health systems, technology developers, patient advocates, and academic institutions, dedicated to promoting responsible AI practices in healthcare.

Buzzworthy developments of the past few days. 

• Healthcare AI proponents say the technology will lower healthcare costs. The assertion has its detractors. One is a senior fellow at the Manhattan Institute. Writing in the think tank’s City Journal, Chris Pope says that, if current reimbursement practices continue treating AI as a nice but unnecessary add-on, care costs will go in the wrong direction. “Medicare’s standard payments already provide adequate incentives to fund new cost-cutting AI systems,” he adds. “If policymakers choose to add on dedicated payments for the use of AI software, tech firms will instead focus on developing cost-hiking innovations, which will only accelerate the upward trend of Medicare expenditures.” Hear him out. 
   
• Picture an AI-ready computer so fast it makes today’s top supercomputers look like, oh, maybe Bell Telephone Laboratories’ Model K Adder from 1937. The K stood for Kitchen, as the tin-can gadget was birthed on its designer’s kitchen table. The blindingly fast descendant of that early computing triumph is Google’s new, experimental quantum computer. As reported by quantum aficionado Cade Metz at the New York Times, the Google machine “needed less than five minutes to perform a mathematical calculation that one of the world’s most powerful supercomputers could not complete in 10 septillion years—a length of time that exceeds the age of the known universe.”
   
• You might not be all that interested in the European Union’s AI Act. But if you’re looking to sell an AI-equipped medical device in the Old Continent, the EU AI Act is going to be interested in you. At Medical Device + Diagnostic Industry, a legal eagle with expertise in the subject offers informal but quite detailed guidance. Even companies without direct regulatory obligations or those operating in non-EU markets “might underestimate the Act’s extraterritorial reach and its implications for global supply chains,” says the expert, Anne-Gabrielle Haie of the international law firm Steptoe. Q&A by MD+DI’s editor-in-chief, Omar Ford. 
   
• Not every healthcare AI tool will succeed. No biggie—‘we need to be OK with some level of failure’ anyway. That’s the stance of Theresa Meadows, senior VP and CIO at Cook Children’s Health Care system in Texas. “We’re trying to figure out the process for balancing innovation with stability, especially in the AI space,” Meadows explains. Get the context behind her comments and more of her thinking at Health System CIO. 
   
• IT leaders tend to overestimate their workers’ readiness for AI. And the workers themselves? Only 17% self-report high levels of skill and comfort with the technology—and well more than a third admit they feel “overwhelmed” by the prospect of having to use AI at work. CIO senior writer Sarah White looks at these and other findings that, taken as a whole, suggest AI can both cause and cure IT burnout. 
   
• Is Google’s ‘everything app’ Astra all that? Yes. No. Maybe. That’s a paraphrase of an AI enthusiast at MIT Technology Review who put the experimental product through its paces. “When it works well, Astra is enthralling,” writes senior editor Will Douglas Heaven. “The experience of striking up a conversation with your phone about whatever you’re pointing it at feels fresh and seamless. This could be generative AI’s killer app.” On the other hand, Astra needs some work. Its reasoning is black box all the way, and it seems to frequently apologize for its mistakes—even when it hasn’t made any. Full review. 
   
• Reddit is out with an AI-powered Q&A interface primarily trained on its users’ posts. Called Reddit Answers, the service is so far only available in English and on a tentative basis while it undergoes fine-tuning. “AI-powered search is part of our longer-term vision to improve the search experience on Reddit—making it faster, smarter and more relevant,” the company explains in a Dec. 9 blog post. Reddit claims 430 million active monthly users, making it the 18th most popular social media platform, according to Exploding Topics. 
   
• It was nice knowing you, Moxie. The charming children’s robot is headed for the great mechanical workshop in the sky. Embodied, the company that has been making Moxie (suggested retail prince $799), says the demise became unavoidable when an expected funding round collapsed while costs of doing business were soaring. There is some chance another company will come to Moxie’s rescue, Embodied says in a FAQs page posted this week. However, if that scenario fails to materialize, it’s lights out. “Moxie relies on cloud connectivity for its core features,” Embodied says, “and it will not function once services end.”
   
• Recent research in the news: 
   
  • Dartmouth: AI thought X-rays of your knees show if you drink beer—they don’t.
     
• Mergers & acquisitions:
   
  • Carta Healthcare to power oncology research and clinical trials with acquisition of Realyze Intelligence, a UPMC Enterprises company 
     
• Funding news of note:
   
  • What we know about XAI as it bolsters itself for AI race—with $6B in new funding
     
  • Eye on AI: Funding to biotech and healthcare startups using AI takes off after dip in 2023
     
• From AIin.Healthcare’s news partners:
   
  • Health Exec: FDA commissioner urges health systems to strengthen AI quality oversight
     
  • Health Imaging: Hospital creates ‘virtual reality technologist’ role to support patients during interventional procedures
     
  • Cardiovascular Business: Cardiologists to highlight the many benefits of AI-powered CCTA evaluations
     
  • Health Imaging: Do large language models help or hinder workflows related to radiology reports?