Page 56 - Read Online

Page 459 Goyal et al. Art Int Surg. 2025;5:450-64 https://dx.doi.org/10.20517/ais.2024.108

Table 5. Newcastle-Ottawa Scale (NOS) for risk of bias assessment
Authors Reviewers Selection Comparability Outcomes Total Final score after consensus
Ortenzi et al. (2023) [10] 1st 4 2 2 8 8
2nd 4 1 2 7
3rd 4 2 2 8
[20]
Takeuchi et al. (2023) 1st 3 1 3 7 7
2nd 4 1 3 8
3rd 3 0 1 4
[21]
Choksi et al. (2023) 1st 3 1 2 6 6
2nd 3 2 3 8
3rd 3 2 1 6
Takeuchi et al. (2022) [8] 1st 3 1 2 6 6
2nd 4 2 2 8
3rd 4 1 1 6
[9]
Zang et al. (2023) 1st 3 1 2 6 6
2nd 3 1 2 6
3rd 3 1 2 6
[7]
Zygomalas et al. (2024) 1st 2 1 2 5 5
2nd 2 1 2 5
3rd 2 1 1 4

Bias sources were further analyzed to identify recurring issues across the studies. The most frequent
concerns included attrition bias due to incomplete outcome data and single-center study designs, which
limited the diversity of datasets. Additionally, imbalanced case distributions, where the number of videos
analyzed per surgeon varied, were observed. In some instances, inadequate case annotations impacted the
robustness of outcome assessments. These limitations collectively contribute to the moderate quality ratings
assigned to the majority of studies in this analysis [Table 6].

DISCUSSION
In this systematic review, we evaluated studies utilizing AI and machine learning (ML) techniques to train
models for real-time surgical phase recognition during minimally invasive IHR procedures. To our
knowledge, this is the first systematic review to focus on this critical intersection of AI and surgical
workflow, underscoring its potential to transform the future of surgery. AI holds immense promise for
healthcare, offering opportunities to reduce labor demands and improve the quality of clinical analysis and
decision making . Over the past few years, the role of artificial intelligence in hernia surgery has been
[22]
examined from broad perspectives. For instance, Lima et al. (2024) performed a qualitative systematic
[23]
review of 13 peer-reviewed studies published between 2020 and 2023. Their analysis suggests that machine
learning platforms hold promise as tools for predicting surgical outcomes and identifying risk factors to
mitigate postsurgical complications.

The integration of AI into healthcare is already reshaping patient care across multiple domains. For
instance, AI-based systems like Babylon Health are enhancing both in-person and virtual consultations by
providing timely and accurate assistance . Similarly, AI-powered tools such as Sense.ly, which combines a
[24]
virtual assistant with an intuitive, human-like interface, are advancing medication management and health
[25]
monitoring in AI-driven diagnostics. A particularly notable example of AI-driven diagnostics is Caption
Health’s FDA-approved software, which enables healthcare providers to perform cardiac ultrasound
imaging without extensive sonographic expertise. Acting as a “co-pilot,” this AI software offers real-time
guidance, instructing users on transducer manipulation and ensuring the acquisition of diagnostic-quality
images .
[26]