内容来源：https://news.microsoft.com/source/asia/features/at-aged-care-provider-regis-ai-takes-on-paperwork-so-staff-can-focus-on-residents/

内容总结：

澳大利亚养老院引入AI助手：68页报告3分钟读完，护士回归临床一线

自2025年9月起，澳大利亚Regis养老护理集团在其72家养老院中部署了一款名为RegiCare Assist的人工智能助手，用于处理日常护理报告中冗长的文字记录。这款基于微软Copilot Studio和Microsoft Foundry开发、由合作伙伴Cognizant协助落地的AI工具，能够自动抓取关键临床信息，并按紧急程度、症状类别（如感染迹象、情绪波动、排便情况等）进行分类整理。

在此之前，临床护理经理每天需要花费数小时阅读一线护工和护士撰写的进度笔记和报告。位于墨尔本一家Regis养老院的临床护理经理多卡斯·桑加朗表示，每天早班开始后，她必须先消化前一夜产生的全部居民护理记录，才能主持9:30的晨会。如今，只需将长达68页的24小时报告上传至RegiCare Assist，几分钟内即可生成一份仅3页的摘要，涵盖当日的临床事件全景。

“我不用再盯着电脑做 paperwork（案头工作），而是有更多时间在楼层里走动，坐下来听居民反馈。”桑加朗说。

Regis首席护理官拉米兹·哈桑强调，AI的定位是“辅助而非替代”临床判断。“居民安全是底线，我们明确设定了这一原则。”为确保准确性，系统采用了RAG（检索增强生成）技术，答案仅限于Regis自身制定的临床政策与操作规程，同时通过点击式界面限制用户输入模糊问题，降低安全风险。

Regis首席信息官伊姆蒂亚兹·巴亚特透露，团队在提示词工程上花了大量时间。例如，指令中“列出所有需要照护的居民”中的“所有”一词必不可少——去掉它，系统有时会遗漏部分人员。

目前，约150名员工已使用该助手。尽管尚无法量化评估其对临床治理水平的提升效果，但反馈普遍积极。“压力变小了，我们对事件有了掌控感。”凯恩斯养老院的临床护理经理玛丽安娜·乔治说。

未来，RegiCare Assist将升级为与现有护理管理系统的一体化集成，从而省去手动上传报告步骤。巴亚特表示：“我们花在电脑上的时间越少，留给居民的实际照护时间就越多。”

中文翻译：

多卡斯·桑加朗上早班才半小时，就已经掌握了夜班期间有关她工作的养老院居民提出的临床问题。
她知道在这家拥有97张床位的养老院里，谁摔倒了，谁拒绝服药，以及是否有人接近生命终点。她也为上午9点半的会议做好了准备，在会上她向自己负责的注册护士和个人护理助理团队指出当天的跟进事项和优先工作。
“我的职责就是确保没有任何遗漏，”这位澳大利亚墨尔本瑞吉斯养老院的临床护理经理说。
这项工作包括花费数小时仔细阅读一线护理人员和护士撰写的多页进展记录和报告——这是对像桑加朗这样的经理来说的每日必读内容。
但自2025年9月以来，他们得到了RegiCare Assist的帮助。该工具基于微软Copilot Studio和微软Foundry构建，由瑞吉斯养老院与解决方案合作伙伴Cognizant共同开发。
这个人工智能助手能汇总冗长的记录，标记关切问题，并按临床问题进行分类。
“今天24小时报告有68页长，”位于澳大利亚东北部城市凯恩斯的瑞吉斯养老院临床护理经理玛丽亚玛·乔治说，“我把它上传到RegiCare Assist。几分钟内，我就能打印出一份临床事件摘要，了解整体情况。只有三页。”
缓解这一痛点正是瑞吉斯在2024年首次考虑部署人工智能时的切入点。如今，RegiCare Assist已被其澳大利亚72家养老院网络中约150名员工使用。
“我们意识到人工智能非常擅长吸收大量文本、捕捉情感和提取关键主题，”瑞吉斯首席信息官伊姆蒂亚兹·巴亚特说。
关于居民健康状况的每日更新现已被RegiCare Assist按关键类别分类，包括即时临床关切、临床趋势、躁动、疼痛或感染迹象以及排便情况。
这些信息得到及时整理和呈现，意味着临床护理经理可以花更少时间阅读，更多时间行动。除了监控当天的优先病例并应对临床事件和紧急情况外，经理们还可能要与家属会面、审查新入院档案并进行员工培训。
“不用再专注于文书工作，我有更多时间巡视现场，”桑加朗说，“还能真正坐下来和居民交流，听取他们的反馈。”
让这些一线经理参与进来对开发RegiCare Assist至关重要，这不仅是为了获得他们的支持，也是为了确保系统不会危及居民安全。
“我们告诉他们：这个（人工智能）是辅助你的，而不是取代你的临床判断或决策，”瑞吉斯首席护理官拉米兹·哈桑说，“我们明确设定了这些参数，因为居民安全至关重要。”
瑞吉斯使用微软Copilot Studio（一个用于构建和定制基于人工智能的代理的低代码平台）来设计围绕日常临床场景的对话流程，并优化面向人工智能助手的提示词，该助手由微软Foundry中的大语言模型提供支持。
为提高准确性和安全性，人工智能助手采用了RAG（检索增强生成）技术，旨在从瑞吉斯临床政策和程序的知识库中生成信息。同时，一个基于点击操作并配有经批准的提示词的界面，减少了模棱两可问题和不安全回答的风险。
“我们花费了大量时间来设计提示词，让用户使用起来更简单，”巴亚特说。他们还花时间构建提示词结构，以确保居民不会在人工智能的回答中被遗漏。
“有一个例子是，当我们说‘总结所有需要关注需求的居民’时。我们发现，如果前面加上‘所有’这个词，它就会涵盖每一个人。而如果没有‘所有’这个词，有时意味着有些人会被忽略，”他解释道。
巴亚特表示，瑞吉斯选择微软的人工智能解决方案，是因为其安全性和易用性。为保护居民的个人信息，人工智能助手在瑞吉斯的安全环境内运行，对数据的访问和共享有严格控制。
巴亚特也对这项技术的快速演进印象深刻。
“我们看到Copilot从发布以来发展非常迅速，”他说。
RegiCare Assist未来将迎来审查和升级。首要任务是与护理提供者现有的护理管理系统实现集成解决方案，这样经理们就不再需要手动上传24小时报告。
首席护理官哈桑表示，现在判断人工智能助手是否改善了瑞吉斯的临床管理还为时过早。但迄今为止的反馈是积极的。
“我们确实有一些来自使用过它的临床医生的轶事数据和反馈，说明它如何减少了时间并帮助他们。当然，任何节省下来的时间……都让他们有更多时间专注于居民的临床护理，”他说。
“我们花在电脑上的时间越少，就能有越多时间与居民在一起，思考如何照顾好他们，”巴亚特说。
到目前为止，临床护理经理桑加朗和乔治都认为，他们的人工智能助手降低了焦虑水平，增强了信心。
“我们能够掌控各种事件，不再为那24小时报告而紧张了，”乔治说，“现在我们的记录处于可控水平。”
顶部图片：瑞吉斯临床护理经理多卡斯·桑加朗与一位养老院居民在一起。自从使用RegiCare Assist后，她有更多时间坐下来与居民交流，听取他们的反馈。摄影：格雷厄姆·丹霍姆，为微软拍摄。
林爱琳为微软消息源报道人工智能相关话题，重点关注人工智能如何改善亚洲地区人们的生活。爱琳曾担任新加坡《海峡时报》副外事编辑，偶尔仍会撰写周末专栏。可通过领英与她联系。
瑞吉斯首席信息官分享关于带领团队踏上人工智能之旅的建议
从你的同事已经感受到的痛点入手。这是瑞吉斯养老院首席信息官伊姆蒂亚兹·巴亚特在成功向澳大利亚各地养老院的约150名员工推出人工智能助手后给出的建议。阅读他关于如何在人工智能之旅中带动员工的更多建议。

英文来源：

Dorkas Sangalang is just 30 minutes into her morning shift, and she’s already up to speed on clinical concerns raised overnight about the residents of the aged care home she works at.
She knows who in the 97-bed home has had a fall, who’s refusing medication and if someone is nearing end of life. And she’s all set for the 9:30 a.m. meeting, where she flags the day’s follow-ups and priorities to her team of registered nurses and personal care assistants.
“My role is to make sure that nothing is missed,” said the clinical care manager of a Regis Aged Care home in Melbourne, Australia.
This involves spending hours scouring pages of progress notes and reports written by frontline carers and nurses – required daily reading for managers like Sangalang.
But since September 2025, they’ve had help from RegiCare Assist, built with Microsoft Copilot Studio and Microsoft Foundry and developed by Regis Aged Care with solutions partner Cognizant.
This AI assistant summarizes the voluminous notes, flags concerns and sorts them by clinical issue.
“Today’s 24-hour report is 68 pages long,” said Mariamma George, a clinical care manager in a Regis home in Cairns, a city in northeast Australia. “I uploaded it into RegiCare Assist. Within minutes, I could print out a summary of clinical events to get a whole picture of what’s going on. It’s three pages.”
Easing this pain point was what Regis landed on when it first began thinking about deploying AI in 2024. Now RegiCare Assist is used by about 150 employees across its network of 72 aged care homes in Australia.
“We realized that AI was really good at absorbing large blocks of text, picking up sentiment and picking up key themes,” said Regis’ chief information officer Imtiaz Bhayat.
Daily updates about the residents’ health within the progress notes are now sorted by RegiCare Assist into key categories like immediate clinical concerns, clinical trends, agitation, signs of pain or infection, and bowel movements.
Having this information organized and surfaced early means clinical care managers can spend less time reading and more time doing. On top of monitoring that day’s priority cases and responding to clinical events and emergencies, managers may be meeting with family members, reviewing new admission files and conducting staff training.
“Instead of focusing on paperwork, I have more time to walk the floor,” said Sangalang. “And to actually sit down with residents and listen to their feedback.”
Involving these frontline managers was key to developing RegiCare Assist, not just to get their buy-in but also to ensure the system does not compromise resident safety.
“We told them: This (AI) is to support you, not replace your clinical judgment or decision-making,” said Rameez Hassan, chief nursing officer at Regis. “We put those parameters clearly in place because resident safety is critical.”
Regis used Microsoft Copilot Studio, a low‑code platform for building and customizing AI‑powered agents, to design conversation flows around everyday clinical scenarios and refine prompts for the AI assistant, powered by large language models in Microsoft Foundry.
To boost accuracy and safety, the AI assistant uses RAG (retrieval augmented generation) that’s designed to produce information from a knowledge base of Regis’ clinical policies and procedures. Meanwhile, a click-based interface with approved prompts reduces the risk of ambiguous questions and unsafe answers.
“We spent a lot of time engineering the prompts, making it easier for the users,” said Bhayat. They also spent time structuring prompts to ensure residents would not go missing in the AI’s responses.
“There was one instance when we said: Summarize all the residents who have needs to be addressed. We discovered that if you put the word ‘all’ in front, it captures everybody. Whereas not having that word ‘all’ sometimes meant that people dropped off,” he explained.
Regis chose Microsoft’s AI solutions, said Bhayat, because of its security and ease of use. To protect residents’ personal information, the AI assistant runs within Regis’ secure environment, with strict controls on how data is accessed and shared.
Bhayat was also impressed by how quickly the technology was moving.
“We’ve seen Copilot evolving rapidly from the time it was released,” he said.
Future reviews and upgrades are in store for RegiCare Assist. At the top of the list is an integrated solution with the care provider’s existing care management system, so that managers no longer need to manually upload the 24-hour report.
It’s still too early to know whether the AI assistant has improved Regis’ clinical governance, said chief nursing officer Hassan. But the feedback so far is positive.
“We do have some anecdotal data as well as feedback from our clinicians who have been using it in terms of how it has reduced the time and helped them. And of course, any reduced time … gives them more time to focus on resident clinical care,” he said.
“The less time we spend on a computer, the more time we can spend thinking about the care of the resident, with them,” said Bhayat.
So far, clinical care managers Sangalang and George both agree their AI assistant has lowered their anxiety levels and boosted their confidence.
“We are on top of incidents and not stressing about the 24-hour report,” said George. “Now our notes are at a manageable level.”
Top image: Regis clinical care manager Dorkas Sangalang with a care home resident. Since using RegiCare Assist, she has more time to sit down with residents and listen to their feedback. Photo by Graham Denholm for Microsoft.
Lim Ai Leen reports on AI for Microsoft Source, focusing on how it’s improving lives in Asia. Ai Leen was formerly associate foreign editor at The Straits Times in Singapore and still pens an occasional weekend column. Contact her on LinkedIn.
Regis CIO shares his tips on bringing people along on the AI journey
Start with a pain point your colleagues are already feeling. This is what Regis Aged Care chief information officer Imtiaz Bhayat advises, after successfully rolling out an AI assistant to some 150 staff across care homes in Australia. Read more tips from him on how to bring workers along on your AI journey.