Page 113 - Read Online
P. 113

Page 155                                                              Shen et al. Art Int Surg. 2025;5:150-9  https://dx.doi.org/10.20517/ais.2024.71

               cost reduction due to a drastic decrease in the staffing burden required for frequent flap checks, especially
                                             [32]
               within the first 24 h post-operation .
               Predictive analytics can also be utilized to create personalized postoperative care plans and enhance patient-
                                                [33]
               specific, value-based care and recovery . ML models based on preoperative and intraoperative information
               can help plastic surgeons quickly stratify risk factors of flap failure in microvascular breast reconstruction
                                                     [34]
               and guide postoperative treatment planning . For example, ML models have been developed to predict
               30-day reoperation and readmission for head and neck free flap patients using patient demographics and
               perioperative factors (operation time, hospital length of stay, surgical site infection) . In the instances of
                                                                                       [35]
               autologous breast reconstruction, such models could potentially shorten hospital stays and reduce spending
               among low-risk patients while informing preventative measures to reduce flap failure among high-risk
               patients.  For implant-based breast reconstruction, ML models were also able to identify significant
               predictors of periprosthetic infection and explantation, as well as nipple-areolar complex necrosis following
               immediate breast reconstruction [36,37] . Breast reconstruction has been among the top 5 most performed
               reconstructive procedures for the last 2 years, and this volume only continues to grow . As more data
                                                                                           [38]
               become available to train these models, their acuity and surgical utility will continue to improve.
               Additionally, this growing volume of autologous procedures also signals the need for more efficient and
                                                                                [39]
               cost-effective postoperative management strategies, potentially enabled by AI . As we transition to a future
               with shorter hospital stays and more focus on personalized medicine [40,41] , AI could greatly enhance the
               precision and effectiveness of postoperative care, leading to better patient outcomes and overall satisfaction.


               Other AI applications that are revolutionizing postoperative care include remote monitoring and
               telemedicine. Wearable AI devices have been applied in aesthetic postoperative settings to monitor patient
                                                    [42]
               compliance with positioning after discharge . This could be used to detect clinical derangements and alert
               surgeons in real time, enabling prompt response and better outcomes, while allowing patients to recover
               from the comfort of their own homes.

               Patient satisfaction and outcomes initiatives
               Patient satisfaction with breast reconstruction following mastectomy is highly variable and subject to
               influence from various factors involved in their cancer treatment process. ML models developed from an
               international  cohort  of  patients  diagnosed  with  breast  cancer  who  underwent  mastectomy  and
               reconstruction demonstrated the ability to accurately predict individual patient-reported outcome measures
                                                                                           [43]
               (PROMs), as evaluated by BREAST-Q, prior to the initiation of the reconstruction process . By integrating
               these AI models into treatment planning for each patient, plastic surgeons may better guide the patients in
               making treatment plans that maximize patient satisfaction.


               AI tools such as chatbots and virtual assistants can significantly enhance the patient experience with
               consultation and education by promptly addressing patients’ queries. For real-time patient question
               resolution, a survey from a single institution demonstrated that 96% of patients were satisfied or somewhat
               satisfied with their interaction with the patient-facing AI chatbots . Regarding general simplification of
                                                                        [44]
               patient instructions and educational materials, Chat Generative Pretrained Transformer 3.5 (OpenAI, San
               Francisco, California, 2022) could effectively simplify these resources down to a broadly acceptable reading
               level . This is especially relevant for patients considering implant-based reconstruction, who often face
                   [45]
               lengthy and evolving educational materials about implant safety and risks, as the Food and Drug
               Administration (FDA) continues to update its recommendations and warnings surrounding breast
               implants . For example, AI-driven tools can help break down lengthy patient materials regarding the safety
                      [46]
               profile and risks associated with these implants into more digestible formats, which will help patients make
   108   109   110   111   112   113   114   115   116   117   118