The telltale signs of infection appeared soon after the patient arrived in the emergency room. So, doctors at China Medical University Hospital (CMUH) ordered a lab test to identify the culprit. Three days later, they got their answer: a drug-resistant bacteria. This particular bug had evolved to outfox most antibiotics, even ones used as a last resort. It is a notorious and potentially fatal threat that haunts hospitals worldwide.
But by then the patient was about to be discharged. That’s because an Artificial Intelligence (AI) model developed by CMUH had already identified the bacteria—in 1 hour, compared with the 72 hours the standard lab test took. Doctors prescribed new drugs as a result, and the patient recovered.
That AI model is a key pillar of CMUH’s “intelligent antimicrobial system,” which was deployed into clinical practice last year. Since then, CMUH has seen patient mortality fall by 25%, antibiotic costs by 30% and antibiotic use by 50%. It has helped doctors treat thousands of patients much like the one who left the ER in March.
And it is just one example of how CMUH is using AI to create a new era of healthcare, said Dr Kai-Cheng Hsu, Director of CMUH’s Artificial Intelligence Centre for Medical Diagnosis.
CMUH has developed and deployed hundreds of AI algorithms, hosted on Microsoft’s Azure cloud platform, which are used every day across the system’s 12 hospitals. The team’s custom-built AI models are helping doctors diagnose diseases like cancer and Parkinson’s. They are helping ER staff treat stroke and heart attack patients more quicker. And they are helping ease the paperwork load on doctors and nurses.
“The major goals of every AI tool we develop are to save patients’ lives and save doctors’ time,” Dr Hsu said. “Our vision is to be a world-class platform provider for the AI hospital of the future.”
Saving Time, Saving Lives
CMUH has been laying the foundations for that AI hospital for years. In 2015 the system established a centre for big data, which could ensure patient data privacy while supplying training fuel for AI algorithms. The AI Center for Medical Diagnosis launched two years later.
Today, 10 different AI clinics operate at CMUH. A cross-disciplinary team of doctors and AI researchers meets monthly to discuss potential new projects and review progress on existing research. Models that are successfully validated and tested are integrated into clinical practice. To date, eight of its AI models have received regulatory approval from Taiwan’s Food and Drug Administration, and eight more have been submitted for review. All have been developed and deployed using the Azure Machine Learning platform.
“Azure is at the core of our digital healthcare transformation,” said Dr Jiaxin Yu, director of CMUH’s AI Innovation Center. “The platform lets us quickly develop new AI tools, deploy them into the clinic and improve the daily practice of our physicians.”
The AI models are often incorporated into familiar software doctors use every day. Some are deployed with a literal push of a button. For example, doctors who order an MRI on a knee injury can click a button marked “AI” that predicts the likelihood of a meniscus tear. That instant result can avoid delays in a follow-up appointment and get a patient treated more quicker.
Saving time often means saving lives. That is increasingly true when it comes to the rapid identification of deadly, drug-resistant bacteria. The World Health Organization predicts that by 2050, antibiotic resistance will cause more deaths than cancer.
But as the patient with drug-resistant bacteria found, the standard lab method can take up to three days. So, CMUH developed a new model. It combined Machine Learning with a common and much simpler lab technique that identifies molecular fingerprints of different bacterial strains. Data on thousands of clinical samples were used to train the algorithm, which can accurately identify six different types of multi-drug resistant bacteria in one hour, according to Dr Yu.
With possible heart attack patients, critical care can be delayed while waiting for a specialist to review an electrocardiogram (ECG), Hsu said. CMUH developed an algorithm that can analyse ECGs, detect the likelihood of a heart attack and send a message to an on-duty specialist for confirmation.
CMUH emergency room staff have used the AI model for two years. It has cut in half the time between when a patient arrives and when they are treated, Hsu said. The algorithm was so successful they decided to partner with a medical device manufacturer to bring it into ambulances as an early warning system.
Fire departments in the cities of Taichung and Nantou have used the system for nearly a year, Dr Hsu said. An ECG can be taken in an ambulance and then uploaded to the cloud, where the AI model can help with a diagnosis and alert medical teams at the hospital to prepare for a heart attack patient. This can cut the time between symptom onset and the start of treatment from two hours down to 10 minutes, he said.
Not all AI models have been as successful, Dr Hsu and Dr Yu clarified.
CMUH developed an algorithm to predict sepsis, a serious illness triggered by infection. Many patients with sepsis show signs of organ failure. When the AI model was trained, it learned to correlate those symptoms with the illness. But not all patients with organ failure have sepsis. As a result, the model incorrectly flagged some patients with organ failure as having sepsis. To solve the problem, Dr Yu said, the team could train future models with additional labels to help it distinguish cases where organ failure is not caused by infection.
Dr Hsu and Dr Yu noted that doctors have always had to make complex decisions using the data they had. AI is there to help support those decisions by providing deeper insights into data, not to make decisions best left to medical experts.
Based on CMUH’s journey, AI can help improve the care doctors deliver, they said. That is why CMUH hopes to share its AI tools broadly using the Azure platform.
“We want to deploy our AI models to hospitals and clinics around the world, and it is much more efficient and cost-effective to use a cloud platform,” Yu said. “Patient privacy is always a top concern, and a critical reason we chose Azure is because it can ensure the safety of sensitive health data. Azure really reduces the barriers to bringing AI into the hospital.”
Archive
- October 2024(27)
- September 2024(94)
- August 2024(100)
- July 2024(99)
- June 2024(126)
- May 2024(155)
- April 2024(123)
- March 2024(112)
- February 2024(109)
- January 2024(95)
- December 2023(56)
- November 2023(86)
- October 2023(97)
- September 2023(89)
- August 2023(101)
- July 2023(104)
- June 2023(113)
- May 2023(103)
- April 2023(93)
- March 2023(129)
- February 2023(77)
- January 2023(91)
- December 2022(90)
- November 2022(125)
- October 2022(117)
- September 2022(137)
- August 2022(119)
- July 2022(99)
- June 2022(128)
- May 2022(112)
- April 2022(108)
- March 2022(121)
- February 2022(93)
- January 2022(110)
- December 2021(92)
- November 2021(107)
- October 2021(101)
- September 2021(81)
- August 2021(74)
- July 2021(78)
- June 2021(92)
- May 2021(67)
- April 2021(79)
- March 2021(79)
- February 2021(58)
- January 2021(55)
- December 2020(56)
- November 2020(59)
- October 2020(78)
- September 2020(72)
- August 2020(64)
- July 2020(71)
- June 2020(74)
- May 2020(50)
- April 2020(71)
- March 2020(71)
- February 2020(58)
- January 2020(62)
- December 2019(57)
- November 2019(64)
- October 2019(25)
- September 2019(24)
- August 2019(14)
- July 2019(23)
- June 2019(54)
- May 2019(82)
- April 2019(76)
- March 2019(71)
- February 2019(67)
- January 2019(75)
- December 2018(44)
- November 2018(47)
- October 2018(74)
- September 2018(54)
- August 2018(61)
- July 2018(72)
- June 2018(62)
- May 2018(62)
- April 2018(73)
- March 2018(76)
- February 2018(8)
- January 2018(7)
- December 2017(6)
- November 2017(8)
- October 2017(3)
- September 2017(4)
- August 2017(4)
- July 2017(2)
- June 2017(5)
- May 2017(6)
- April 2017(11)
- March 2017(8)
- February 2017(16)
- January 2017(10)
- December 2016(12)
- November 2016(20)
- October 2016(7)
- September 2016(102)
- August 2016(168)
- July 2016(141)
- June 2016(149)
- May 2016(117)
- April 2016(59)
- March 2016(85)
- February 2016(153)
- December 2015(150)