Scholarlink

Background <p>Artificial intelligence (AI) and machine learning (ML) are increasingly integrated into metabolic and bariatric surgery (MBS), offering opportunities to enhance decision-making, optimize perioperative care, and personalize outcomes. However, their clinical adoption requires a comprehensive understanding of current applications and limitations.</p> Objectives <p>To systematically review the use of AI and ML across the bariatric surgery pathway, from preoperative planning to postoperative follow-up.</p> Methods <p>A systematic search of PubMed and Embase (last accessed August 28, 2025) identified original studies applying AI/ML in MBS. Inclusion criteria were studies involving bariatric patients and AI/ML-based models for clinical or perioperative purposes. Data on study design, sample size, surgical procedure, AI/ML technique, and primary outcomes were extracted. Studies were categorized into preoperative, intraoperative, and postoperative phases.</p> Results <p>Of 142 records screened, 27 studies met inclusion criteria. Preoperative applications focused on patient selection, anatomical prediction, and risk stratification, including models predicting weight-loss success, gastroesophageal reflux disease, and hiatal hernia. Intraoperative research explored operative time forecasting, workflow optimization, and automated video analysis for skill assessment and step recognition. Postoperative models addressed complication prediction, nutritional surveillance, and long-term weight trajectory forecasting. Reported accuracy was high (AUC up to 0.93), but external validation and fairness audits were limited. Emerging evidence also supports AI-driven educational tools, including large language models for surgical training.</p> Conclusions <p>AI is rapidly transforming bariatric surgery across the preoperative, intraoperative, and postoperative pathway, offering unprecedented opportunities for personalization, efficiency, and quality improvement. From risk prediction to skill assessment and long-term outcome forecasting, its applications can augment, rather than replace, human expertise. Ethical safeguards, transparency, and equitable access remain critical for safe and effective integration.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Artificial intelligence and machine learning in bariatric surgery: a comprehensive systematic review

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月1日

查看全文
  • Antonio Vitiello,
  • Giovanna Berardi,
  • Maria Spagnuolo,
  • Roberto Peltrini,
  • Vincenzo Pilone

摘要

Background

Artificial intelligence (AI) and machine learning (ML) are increasingly integrated into metabolic and bariatric surgery (MBS), offering opportunities to enhance decision-making, optimize perioperative care, and personalize outcomes. However, their clinical adoption requires a comprehensive understanding of current applications and limitations.

Objectives

To systematically review the use of AI and ML across the bariatric surgery pathway, from preoperative planning to postoperative follow-up.

Methods

A systematic search of PubMed and Embase (last accessed August 28, 2025) identified original studies applying AI/ML in MBS. Inclusion criteria were studies involving bariatric patients and AI/ML-based models for clinical or perioperative purposes. Data on study design, sample size, surgical procedure, AI/ML technique, and primary outcomes were extracted. Studies were categorized into preoperative, intraoperative, and postoperative phases.

Results

Of 142 records screened, 27 studies met inclusion criteria. Preoperative applications focused on patient selection, anatomical prediction, and risk stratification, including models predicting weight-loss success, gastroesophageal reflux disease, and hiatal hernia. Intraoperative research explored operative time forecasting, workflow optimization, and automated video analysis for skill assessment and step recognition. Postoperative models addressed complication prediction, nutritional surveillance, and long-term weight trajectory forecasting. Reported accuracy was high (AUC up to 0.93), but external validation and fairness audits were limited. Emerging evidence also supports AI-driven educational tools, including large language models for surgical training.

Conclusions

AI is rapidly transforming bariatric surgery across the preoperative, intraoperative, and postoperative pathway, offering unprecedented opportunities for personalization, efficiency, and quality improvement. From risk prediction to skill assessment and long-term outcome forecasting, its applications can augment, rather than replace, human expertise. Ethical safeguards, transparency, and equitable access remain critical for safe and effective integration.