전 세계적으로 증가하는 지방간 질환은 초기 증상이 없어 심각한 합병증으로 이어지기 쉽습니다. 의료진의 과중한 업무 부담 속에서 AI가 기존 건강 기록과 정기 혈액 검사 데이터를 자동 분석하여 고위험군을 선별하고 초기 치료를 돕는 핵심 역할을 할 수 있습니다.
번역된 본문
전 세계적으로 10억 명 이상의 사람들에게서 간의 구성에 느리고 은밀한 변화가 일어나고 있다. 정상적이고 건강한 간에는 지방이 거의 존재하지 않지만, 많은 성인과 심지어 어린이들의 간에서 지방이 장기 총 무게의 5% 심지어 10%를 초과하는 것으로 나타났다. 이러한 비정상적인 지방 축적은 염증, 세포 손상, 그리고 섬유화(fibrosis)로 알려진 반흔 조직의 형성을 유발한다. 이는 현재 전 세계 성인의 약 30%에 영향을 미치는 지방간 질환의 전형적인 특징이다. 이를 방치하면 점진적인 지방 축적으로 인해 결국 간 기능 상실로 이어질 수 있으며, 심혈관계 질환 및 다양한 암의 위험 증가와도 연관이 있는 것으로 밝혀졌다. 하지만 이 질환은 뚜렷한 증상 없이 진행되는 경우가 많아 치료가 가능한 초기 단계에서 발견되기는 매우 드물다. 간경변증이나 간의 진행성 반흔(Scarring) 사례의 경우에도, 4분의 3의 환자가 생명을 위협받는 단계가 되어서야 진단을 받는다.
그 결과, 점점 더 많은 전문가들이 AI가 어떻게 도움을 줄 수 있을지 조사하고 있다. 뉴욕시립대학교(CUNY) 대학원 공중보건 및 보건정책학과의 제프리 라자루스(Jeffrey Lazarus) 교수는 AI 도구를 활용해 방대한 전자 건강 기록을 분석하고, 간에 우려할 만한 수준의 지방이 축적되었을 가능성이 가장 높은 사람을 정확히 찾아낼 수 있다고 제안한다. 라자루스 교수는 "AI는 과거의 수많은 병원 방문 기록과 검사 보고서를 돌아볼 수 있습니다. 이를 통해 실제로 가장 위험이 높은 사람의 우선순위를 정할 수 있습니다."라고 말한다. 지방간 질환을 조기에 발견한다면, 상당 부분의 손상은 회복 가능성이 매우 높다. 초기 단계에서는 음주량 감소, 식이 개선 및 운동을 통한 체중 감량, 심지어 커피를 더 많이 마시는 것과 같은 생활 습관 변화만으로도 반흔과 염증을 역전시킬 수 있는 것으로 나타났다. 중등도에서 진행성 간 반흔이 있는 사람들의 경우에도, GLP-1 약물인 세마글루타이드(semaglutide)와 레즈메티롬(resmetirom)이라는 약물과 같은 새로운 치료법이 매우 효과적인 치료제로 입증되었다.
라자루스 교수는 "간은 재생할 수 있고 섬유화를 되돌릴 수 있으며 완전히 다시 건강해질 수 있기 때문에 매우 다재다능하고 흥미로운 장기입니다. 하지만 전통적으로 우리는 환자를 조기에 발견해 질환의 진행을 예방하기보다는, 말기 치료에 더 집중하고 환자를 얼마나 더 오래 살릴 수 있을지에 초점을 맞춰왔습니다."라고 말한다. 라자루스 교수와 다른 전문가들이 좌절감을 느끼는 이유 중 하나는, 간 건강을 평가하는 간단하고 비침습적인 방법이 이미 존재함에도 불구하고 이를 거의 사용하지 않는다는 점이다. 특히 비만이나 제2형 당뇨병이 있는 사람들처럼 지방간 질환의 결과에 따른 위험이 훨씬 더 높은 것으로 알려진 환자들조차도 이러한 평가를 생략하는 경우가 많다. 예를 들어, Fib-4 지수는 환자의 나이, 두 가지 간 효소 수치, 그리고 혈액 응고 능력을 기반으로 0에서 6 사이의 점수를 계산하여 진행성 간 섬유화의 위험도를 평가한다. 이를 위해서는 의사가 간 혈액 검사를 실시해야 하며, 이는 미국에서 연례 건강 검진의 일부로 자주 수행된다. 또한 의사들은 반흔 조직을 생성하는 데 관여하는 두 가지 단백질 수치와 간 내 반흔 제거를 억제하는 효소 수치를 측정하는 ELF(Enhanced Liver Fibrosis) 검사라는 보다 정확한 2차 혈액 검사에도 접근할 수 있다. 간 지방 수치가 우려되는 환자에게 이 두 가지 검사를 모두 사용하면 진행성 섬유화 진단율을 4배 향상시킬 수 있는 것으로 나타났다.
하지만 점점 늘어나는 업무와 행정적 부담에 직면한 의사들에게 업무 흐름에 단순히 더 많은 검사를 추가하는 것은 지속 가능한 선택지로 여겨지지 않는다. 예일 대학교 의과대학의 조나단 드라노프(Jonathan Dranoff) 교수는 "백그라운드에서 자동으로 실행되거나 버튼 하나만 누르면 쉽게 처리할 수 있는 시스템이 있어야 합니다."라고 말한다. 그 결과, 드라노프 교수와 라자루스 교수는 모두 AI가 정기적인 혈액 검사의 기존 데이터를 활용하여 Fib-4 점수 계산을 자동화하는 역할을 할 것으로 예상한다.
Comment Loader Save Story Save this story Comment Loader Save Story Save this story All over the world, a slow, insidious change is taking place in the composition of the livers of more than a billion people. While the presence of fat in a normal, healthy liver is negligible, many adults and even children have livers where fat exceeds 5 percent or even 10 percent of the organ’s total weight. Its unnatural presence causes inflammation, cell damage, and the formation of scar tissue known as fibrosis, all hallmarks of fatty liver disease, a condition that now impacts approximately 30 percent of adults worldwide . If left to linger, progressive fat accumulation can ultimately lead to liver failure, and it has been linked to an increased risk of cardiovascular disease and various cancers. But the condition typically develops without noticeable symptoms, so it is rarely detected at an early and treatable stage. Even in cases of cirrhosis or advanced scarring of the liver, three-quarters of people are only diagnosed once their condition has become life-threatening. As a result, growing numbers of specialists are investigating how AI might help. Jeffrey Lazarus, a professor at the CUNY Graduate School of Public Health and Health Policy, suggests that AI tools could be used to comb through vast numbers of electronic health records and pinpoint the people most likely to have accrued worrying amounts of liver fat. “AI can retrospectively go through massive numbers of hospital visits and lab reports,” says Lazarus. “You can use that to really prioritize who’s at most risk.” If fatty liver disease can be identified early, much of the damage is highly reversible. In the initial stages, lifestyle changes such as reducing alcohol intake, losing weight through dietary improvements and exercise, and even drinking more coffee have all been shown to be capable of reversing scarring and inflammation. Even for people with moderate to advanced liver scarring, novel treatments such as the GLP-1 medication semaglutide and a drug called resmetirom have been shown to be highly effective therapeutics. “The liver is a very versatile and interesting organ because it can regenerate, fibrosis can be reversed, and you can be completely healthy again,” says Lazarus. “But traditionally, we’ve focused more on late-stage care and trying to see how long we can keep a patient alive rather than finding them early and preventing the condition from advancing.” One of the frustrations for Lazarus and others is that while simple, noninvasive means of assessing liver health exist, they are rarely used, even in people known to be at a much higher risk of the consequences of fatty liver disease, such as those with obesity and type 2 diabetes. The Fib-4 index, for example, rates a person’s risk of advanced liver fibrosis through computing a score of between 0 and 6, based on their age, their levels of two liver enzymes, and their blood-clotting ability. This requires physicians to run a liver blood test, something which is often carried out as part of an annual medical checkup in the US. Doctors also have access to a more accurate, second-line blood test known as the enhanced liver fibrosis test, which measures the levels of two proteins involved in creating scar tissue and an enzyme which inhibits the clearance of scars in the liver. Using both of these tests in patients with worrying amounts of liver fat has been shown to improve the diagnosis of those with advanced fibrosis by four-fold. But for physicians facing a growing workload and administrative burden, simply adding more testing to the workflow is not felt to be a sustainable option. “You have to have something that can run in the background or it’s easy to just hit a button and do it,” says Jonathan Dranoff, a professor of medicine at Yale University. As a result, Dranoff and Lazarus both foresee a role for AI in taking existing data from routine blood testing and automating the calculation of Fib-4 scores, making it easier for primary care physicians to identify the right patients to refer to a liver specialist. AI could also do this using x-ray images. Last year, scientists at Osaka Metropolitan University in Japan published a study using an AI model capable of ingesting and analyzing routine chest x-ray scans. While primarily intended to examine the lungs and the heart, these images also pick up parts of the liver. The researchers found that their model was capable of identifying people with fatty liver disease with an accuracy of 82 percent. Lazarus suggests that AI-powered algorithms could be incorporated into all routine x-ray analyses where the liver happens to be scanned alongside other organs. “The AI could pick up cases of excess liver fat, check for other risk factors such as if the person is overweight, has high cholesterol or type 2 diabetes, and then make a recommendation to the doctor,” says Lazarus. “It could tell them, ‘This might not have been what you were looking for, but this is what was picked up, and you should refer to hepatology or endocrinology who can do further tests.’” Training AI algorithms on routine blood tests is also beginning to yield better diagnostics for progressive fatty liver disease. While Fib-4 is a quick and low-cost tool, it is less accurate in certain age groups, such as adolescents and seniors, and studies have raised concerns about rates of false positives and unnecessary referrals unless combined with additional testing. As a result, the Danish health tech startup Evido has developed an AI-powered algorithm called LiverPRO which can assess a patient’s risk of liver fibrosis based on their age and nine routine blood-based biomarkers. Now being commercialized in partnership with the pharmaceutical company Roche, it has been shown to outperform Fib-4 in predicting risk of serious liver problems in more than 470,000 middle-aged people. Other AI-based tests could also help doctors select the most appropriate patients for resmetirom treatment without requiring them to undergo an invasive liver biopsy. Earlier this year, an international collective of hepatologists published the results of a study evaluating an AI model called ALADDIN—another machine learning algorithm based on routine blood tests—that showed it performed better than Fib-4 and other risk scores in identifying the patients who could benefit most from the drug. “These tools won’t completely replace biopsies or imaging,” says Paul Brennan, a specialty registrar in gastroenterology, hepatology, and internal medicine at the University of Dundee. “But they could fix the bottlenecks in primary care where most fibrosis goes undetected. I expect them to be adopted as a smarter first pass, catching the moderate-risk patients that blunter tools may miss, and reducing unnecessary referrals to hepatologists.” So far, the use of AI in liver care has largely been confined to research, but Lazarus is optimistic that this will start to change soon. He points to research carried out in Denmark which found that informing people that they have liver fibrosis makes them more likely to commit to dietary and exercise regimes. “We’re always looking to improve adherence to eating well and doing more physical activity,” says Lazarus. “Telling someone that they might have or do have liver disease is one way to do that.” For health care systems grappling with a rising burden of chronic disease, identifying and treating patients in the earlier stages of fatty liver disease would save vast amounts of money. “I would love to say, let’s go back through the electronic medical records across the various US health systems and find people before they have cirrhosis,” says Lazarus. “Liver transplants are extraordinarily expensive in any country, but especially the US. So there’s a lot of good humane and economic reasons to find people earlier on.”