Montemurro Paolo, Lehnhardt Marcus, Behr Björn, Wallner Christoph
Akademikliniken, Storängsvägen 10, 11541, Stockholm, Sweden.
Department of Plastic Surgery, BG University Hospital Bergmannsheil, Ruhr University Bochum, Bürkle-de-la-Camp Platz 1, 44789, Bochum, Germany.
Aesthetic Plast Surg. 2022 Dec;46(6):2669-2676. doi: 10.1007/s00266-022-02997-2. Epub 2022 Jul 8.
Primary breast augmentation is one of the most commonly requested aesthetic procedures. Considering the large number of procedures performed in connection with a high demand, it is crucial to prevent complications. For this reason, finding and avoiding possible sources of complications is decisive.
Between January 2010 and December 2021, 1625 female patients underwent an aesthetic breast augmentation performed by a single surgeon. The data collected were analyzed through a machine learning technique for binary recursive partitioning. This made it possible to detect unknown sources of a complication and determine a vertex for the various features.
When analyzing the data, for most features a high importance score with low entropy was achieved, concluding a high significance. In addition, reproducibility was demonstrated through detailed testing and training accuracies in the algorithm. With this procedure, in addition to known risks such as a high BMI and round implant shape, a larger than A preoperative bra-cup size (OR: 2.7) and a taller body could also be identified as most significant influencing factors for complications.
Preoperative breast size plays an exceptionally important role in the occurrence of complications and should be a factor held in a surgeon's considerations. In addition, this study shows ways to transfer artificial intelligence into plastic surgery to increase medical quality.
This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .
一期隆乳术是最常被要求实施的美容手术之一。鉴于大量手术是在高需求的背景下进行的，预防并发症至关重要。因此，找出并避免可能的并发症来源具有决定性意义。
2010年1月至2021年12月期间，1625名女性患者接受了由同一位外科医生实施的美容隆乳术。收集的数据通过用于二元递归划分的机器学习技术进行分析。这使得能够检测到并发症的未知来源，并确定各种特征的一个顶点。
在分析数据时，对于大多数特征，获得了具有低熵的高重要性得分，得出了高度显著性。此外，通过算法中的详细测试和训练准确性证明了可重复性。通过此程序，除了高BMI和圆形植入物形状等已知风险外，术前胸罩罩杯尺寸大于A（比值比：2.7）和身材较高也可被确定为并发症的最显著影响因素。
术前乳房大小在并发症的发生中起着极其重要的作用，应该是外科医生考虑的一个因素。此外，本研究展示了将人工智能应用于整形手术以提高医疗质量的方法。
证据等级IV：本刊要求作者为每篇文章指定一个证据等级。有关这些循证医学评级的完整描述，请参阅目录或在线作者指南www.springer.com/00266 。
