Gelman Ram, Hurvitz Noa, Nesserat Rima, Kolben Yotam, Nachman Dean, Jamil Khurram, Agus Samuel, Asleh Rabea, Amir Offer, Berg Marc, Ilan Yaron
Departments of Medicine, Hadassah Medical Center, and Faculty of Medicine, Hebrew University, Jerusalem, Israel.
Departments of Cardiology, Hadassah Medical Center, and Faculty of Medicine, Hebrew University, Jerusalem, Israel.
Biomed Pharmacother. 2023 May;161:114334. doi: 10.1016/j.biopha.2023.114334. Epub 2023 Mar 9.
Diuretics are a mainstay therapy for congestive heart failure (CHF); however, over one-third of patients develop diuretic resistance. Second-generation artificial intelligence (AI) systems introduce variability into treatment regimens to overcome the compensatory mechanisms underlying the loss of effectiveness of diuretics. This open-labeled, proof-of-concept clinical trial sought to investigate the ability to improve diuretic resistance by implementing algorithm-controlled therapeutic regimens.
Ten CHF patients with diuretic resistance were enrolled in an open-labeled trial where the Altus Care™ app managed diuretics' dosage and administration times. The app provides a personalized therapeutic regimen creating variability in dosages and administration times within pre-defined ranges. Response to therapy was measured by the Kansas City Cardiomyopathy Questionnaire (KCCQ) score, 6-minute walk test (SMW), N-terminal pro-brain natriuretic peptide (NT-proBNP) levels, and renal function.
The second-generation, AI-based, personalized regimen alleviated diuretic resistance. All evaluable patients demonstrated clinical improvement within ten weeks of intervention. A dose reduction (based on a three-week average before and last three weeks of intervention) was achieved in 7/10 patients (70 %, p = 0.042). The KCCQ score improved in 9/10 (90 %, p = 0.002), the SMW improved in 9/9 (100 %, p = 0.006), NT-proBNP was decreased in 7/10 (70 %, p = 0.02), and serum creatinine was decreased in 6/10 (60 %, p = 0.05). The intervention was associated with reduced number of emergency room visits and the number of CHF-associated hospitalizations.
The results support that the randomization of diuretic regimens guided by a second-generation personalized AI algorithm improves the response to diuretic therapy. Prospective controlled studies are needed to confirm these findings.
利尿剂是治疗充血性心力衰竭（CHF）的主要手段；然而，超过三分之一的患者会出现利尿剂抵抗。第二代人工智能（AI）系统在治疗方案中引入变异性，以克服导致利尿剂疗效丧失的代偿机制。这项开放标签的概念验证临床试验旨在研究通过实施算法控制的治疗方案来改善利尿剂抵抗的能力。
10名有利尿剂抵抗的CHF患者参加了一项开放标签试验，其中Altus Care™应用程序管理利尿剂的剂量和给药时间。该应用程序提供个性化治疗方案，在预定义范围内使剂量和给药时间产生变异性。通过堪萨斯城心肌病问卷（KCCQ）评分、6分钟步行试验（SMW）、N端前脑钠肽（NT-proBNP）水平和肾功能来衡量治疗反应。
基于第二代人工智能的个性化方案减轻了利尿剂抵抗。所有可评估的患者在干预10周内均表现出临床改善。7/10的患者（70%，p = 0.042）实现了剂量减少（基于干预前三周和最后三周的平均剂量）。9/10的患者KCCQ评分有所改善（90%，p = 0.002），9/9的患者SMW有所改善（100%，p = 0.006），7/10的患者NT-proBNP降低（70%，p = 0.02），6/10的患者血清肌酐降低（60%，p = 0.05）。该干预与急诊就诊次数和CHF相关住院次数的减少有关。
结果支持由第二代个性化人工智能算法指导的利尿剂方案随机化可改善利尿剂治疗反应。需要进行前瞻性对照研究来证实这些发现。
