Prognostic Factors and Predictive Models for Reproductive Outcomes in Patients with Intrauterine Adhesions: A Review.

The incidence of intrauterine adhesions (IUAs) is rising in tandem with the increasing number of intrauterine surgeries, making them a significant gynecological disorder. IUAs are commonly observed after induced abortions or dilation and curettage procedures. A recent meta-analysis by Munro et al reported that the incidence of IUAs after common intrauterine surgeries ranged from 16% to 28%. 1 IUAs, often referred to as Asherman syndrome in severe cases, are primarily caused by mechanical injury or intrauterine inflammation associated with uterine surgery or infection. IUAs damage the endometrium and reduce its regenerative capacity, resulting in the formation of endometrial fibrous bands uterine cavity. 2 These adhesions distort the uterine cavity and disrupt the basal endometrial layer. The normal area of the endometrium decreases, resulting in insufficient blood supply to the endometrium and reduced receptivity. This altered uterine microenvironment interferes with embryo implantation, leading to miscarriage and infertility. 3 , 4 Identifying factors associated with pregnancy outcomes in patients with IUAs is essential for improving prevention and therapeutic strategies. Although studies have evaluated hysteroscopic images and clinical data to predict reproductive outcomes in patients with IUAs, predictive accuracy after hysteroscopic adhesiolysis remains limited. The IUA classification system shows significant differences in prognosis assessment. The reproductive outcomes that different studies focus on are not consistent. There are various forms of prediction methods, including simple scoring systems and complex artificial intelligence algorithms. However, there is currently no fertility prediction model after TCRA that has been widely clinically verified. This review aimed to summarize factors associated with pregnancy outcomes in patients with IUAs and recent advances in their prediction.

The fibrous tissue of IUAs can partially or completely obliterate the uterine cavity. 5 Therefore, IUAs are associated with reduced menstrual flow, amenorrhea, chronic pelvic pain, infertility, recurrent miscarriage, and abnormal placental implantation. 2 , 6 IUAs are primarily associated with repeated intrauterine procedures, curettage, and postoperative infections. However, the relationships between these factors and IUAs remain incompletely understood. The reported prevalence of IUAs varies across studies because of differences in study populations, diagnostic criteria, and classification systems. Consequently, establishing a universally accepted prevalence estimate remains challenging. 6 Classification of IUAs is essential for guiding treatment selection based on the extent of adhesions and for more accurately predicting reproductive outcomes more accurately. 2 An ideal classification system should describe the extent and type of adhesions, incorporate reproductive history, and help predict pregnancy outcomes. Table 1 summarizes classification systems commonly used worldwide. 7 Table 1 Classification Systems for Asherman Syndrome (Adapted from Doroftei et al 7 ; CC BY 4.0, https://creativecommons.org/licenses/by/4.0/ ) March et al (1978) 8 Mild Filmy, avascular adhesions that are easily lysed. The endometrial cavity is minimally involved Moderate Dense, vascular adhesions that require sharp dissection. The cavity is partially occluded. Severe Very dense, broad adhesions that markedly distort or occlude the uterine cavity. Associated with amenorrhea and poor reproductive outcomes. American Fertility Society (1988) 9 Extent of cavity involved Less than one third 1 One third to two thirds 2 More than two thirds 4 Type of adhesion Filmy 1 Filmy and dense 2 Dense 4 Menstrual pattern Normal 0 Hypomenorrhea 2 Amenorrhea 4 Prognostic stage (HSG and hysteroscopy score) Stage I (mild) 1–4 Stage II (moderate) 5–8 Stage III (severe) 9–12 European Society of Gynecological Endoscopy (1995) 10 I Thin or filmy adhesions easily ruptured by hysteroscope sheath alone. Cornual areas normal. II Singular dense adhesion connecting separate parts of the uterine cavity. Visualization of both tubal ostia possible. Cannot be ruptured by hysteroscope sheath alone. IIa Occluding adhesions only in the region of the internal cervical os. Upper uterine cavity normal. Only to be classified by hysteroscopy. III Multiple dense adhesions connecting separate parts of the uterine cavity. Unilateral obliteration of ostial areas of the tubes. IV Extensive dense adhesion with (partial) occlusion of the uterine cavity. Both tubal ostial areas (partially) occluded. Va Extensive endometrial scarring and fibrosis in combination with grade I or grade II adhesions, with amenorrhea or pronounced hypomenorrhea. Vb Extensive endometrial scarring and fibrosis in combination with grade III or grade IV adhesions, with amenorrhea Nasr et al (2000) 11 Isthmic fibrosis Present 2 Filmy adhesions > 50% of the cavity 1 < 50% of the cavity 2 Dense adhesions Single band 2 Multiple bands 4 Tubal ostium Both visualized 0 Only one visualized 2 Both not visualized 4 Tubular cavity Sound < 6 10 Menstrual pattern Normal 0 Hypomenorrhea 4 Amenorrhea 8 Reproductive performance Good obstetrics history 0 Recurrent pregnancy loss 2 Infertility 4 Prognostic Classification Mild 0–4 Moderate 5–10 Severe 11–22 Note : HSG, hysterosalpingography, a radiographic procedure using contrast medium to assess uterine morphology and fallopian tube patency. Classification Systems for Asherman Syndrome (Adapted from Doroftei et al 7 ; CC BY 4.0, https://creativecommons.org/licenses/by/4.0/ ) Note : HSG, hysterosalpingography, a radiographic procedure using contrast medium to assess uterine morphology and fallopian tube patency. Before hysteroscopy became widely available, cervical dilation and blind uterine exploration were commonly used to treat IUAs. Cervical dilation involves manually dilating the cervix with metal dilators and attempting to separate adhesions with a curette or a blunt instrument. However, this technique lacks precision and carries a substantial risk of uterine perforation or further damage to the endometrium. During blind uterine dilation, the clinician relies solely on tactile feedback without hysteroscopic guidance to explore and dilate the uterine cavity. This approach may cause some adhesions to be missed or healthy endometrium to be inadvertently damaged, thereby increasing the risk of postoperative recurrence. The technique of dilation and curettage is associated with short- and long-term complications, including cervical lacerations, bleeding, infection, and uterine perforation, which can extend to bladder or bowel injury. 12–14 Hysteroscopy has evolved from a diagnostic method to a standard treatment tool for IUAs. Hysteroscopic adhesion release surgery can restore normal menstruation and improve the fertility of patients with IUAs. 15 , 16 Hysteroscopy can be used in combination with various instruments when treating IUAs, including monopolar and bipolar electrocautery, plasma energy equipment, and micro-uterine mirror scissors. 17 Reportedly, the recurrence rate of adhesions is 21.8–41.9% and even higher in severe cases. 18 , 19 Excessive adhesiolysis during transcervical resection of adhesions (TCRA) may damage the endometrium and contribute to postoperative adhesion recurrence. 20 Although electrosurgical resection can precisely remove adhesions and achieve good hemostasis, thermal damage may further harm the endometrium. 3 , 21 The use of cold knives or miniature scissors may minimize endometrial trauma. 22 Adhesion recurrence is a major determinant of postoperative pregnancy rates. 23 Therefore, selecting a surgical approach that minimizes endometrial injury and optimizes reproductive outcomes is essential. Following TCRA, adjuvant therapy typically involves hormonal treatment or placement of an intrauterine physical barrier to reduce the risk of recurrent adhesions. 2 , 24 The use of anti-adhesion barriers has become increasingly common. Clinically available anti-adhesion materials—including solutions, films, powders, and hydrogels—remain associated with prolonged healing times and postoperative adhesion recurrence. 25 , 26 Physical barriers made from natural polymers, such as hyaluronic acid, alginate, and chitosan derivatives, have also been applied to prevent adhesion formation between surgical sites and adjacent tissues. 27 For primary prevention, intrauterine gel barriers substantially reduce postoperative IUA formation, with reported relative risks ranging from 0.29 to 0.45. However, for secondary prevention after TCRA, recurrence rates remains high (28–43%) regardless of adjuvant therapy use, indicating the need for more preventive strategies. 1 Commonly used anti-adhesion measures are summarized in Figure 1 . 28 Figure 1 Recent developments in agents for primary prevention of intrauterine adhesions after uterine surgery (Reproduced from Lee et al 28 ; CC BY 4.0, https://creativecommons.org/licenses/by/4.0/ ). The infographic illustrates various methods for preventing intrauterine adhesions. Absorbable barriers include crosslinked hyaluronic acid (CHA) gel or hyaluronic acid (HA) gel. Physical barriers are depicted with intrauterine balloon catheters featuring a flexible silicon balloon, intrauterine device (IUD), Foley balloon catheter and Foley catheter with a Foley balloon. Hormones shown are estradiol, danazol and misoprostol, with their chemical structures. Cell or bio-agents are represented by bioagents and cell therapy. Each method is visually associated with the uterus to indicate its application in preventing adhesions. Infographic: prevent intrauterine adhesions with barriers, hormones and bio-agents. Recent developments in agents for primary prevention of intrauterine adhesions after uterine surgery (Reproduced from Lee et al 28 ; CC BY 4.0, https://creativecommons.org/licenses/by/4.0/ ). Early and regular hysteroscopic follow-up can improve postoperative pregnancy outcomes in patients with IUAs by restoring uterine cavity morphology, reducing adhesion recurrence, and improving menstrual pattern. 29 For severe IUAs, pregnancy outcomes can be improved by performing a second adhesion release surgery under the guidance of an IUD. 30 Hu et al 31 reported that balloon uterine stent placement after TCRA more effectively promotes endometrial repair and functional recovery. Further studies have evaluated the efficacy of TCRA combined with periodic balloon dilation in treating IUAs and its impact on the reproductive outcomes of women undergoing in vitro fertilization. This method improved the intrauterine environment, increased endometrial thickness, and resulted in higher clinical pregnancy and live birth rates. 32 Studies have reported no significant differences in pregnancy or adhesion recurrence rates between postoperative treatment with balloon stents and IUDs in patients with IUAs. However, balloon stents can better maintain the stability of the uterine microbiota, whereas IUD use may increase bacterial colonization. 33 Zan et al 34 developed a dual-function nanostructure consisting of a silver metal-organic framework coated with hyaluronic acid and embedded within a poly-lactic-co-glycolic acid scaffold. It has antibacterial, anti-inflammatory, and anti-fibrotic effects. Preliminary studies have shown that collagen scaffolds may be more effective when used in cases of moderate adhesions. 35 Traditional stent designs exhibit poor adaptability to the three-dimensional structure of the uterine cavity. 36 , 37 Xu et al 38 developed a “micro cold knife plow-like” surgical technique and a scaffold that conforms to the shape of the uterine cavity. This method restores the uterine cavity structure without causing thermal damage and also reduces the recurrence of adhesions caused by mismatch of the stent shape. Preoperative estrogen therapy has been associated with lower American Fertility Society (AFS) scores and reduced surgical complexity. 39 After adhesion release surgery, a daily estrogen dose of 4 mg was as effective as higher doses (6–8 mg) in restoring the menstrual pattern, reconstructing the uterine cavity, and improving fertility. In contrast, the placement of an intrauterine contraceptive device after surgery has a relatively poor effect in preventing adhesion recurrence, and it may also increase the risk of device entrapment. 40 Feng et al 39 recently developed a scaffold that transforms systemic drug administration into slow estrogen release within the uterus, providing a promising postoperative anti-adhesion option for patients with moderate-to-severe IUAs. Compared with the traditional Foley balloon combined with oral estrogen method, this stent demonstrated a higher non-adhesion rate (44.17% vs 19.82%), thicker endometrium (8.46 ± 7.37 mm vs 4.84 ± 2.48 mm), and better adhesion improvement rate (93.33% vs 58.56%) at 60 days. Concurrently, this controlled-release system reduced adverse reactions caused by systemic administration. Biological gel materials are widely used as anti-adhesive agents after hysteroscopy procedures as they can effectively prevent the recurrence of mild adhesions after surgery and significantly increase the pregnancy rate after miscarriage. However, their effect in preventing the recurrence of severe IUAs remains limited. 41 , 42 They are more suitable for application in the early postoperative period, owing to their relatively fast degradation rate (48–72 hours). 43 Hydrogels have strong tissue adhesion properties and can effectively maintain the internal space of the uterus. An autologous platelet-rich gel can reduce the recurrence rate of adhesions in patients with moderate IUAs and increase the pregnancy rate. 44 Self-healing hydrogels possess antioxidant and adhesive properties and can closely conform to the uterine cavity. 45 , 46 Compared with traditional stem-cell therapy, combining exosomes derived from decidual stromal cells and sodium alginate hydrogel can achieve high stability and low immunogenicity. 47 Because of the common problem of poor adhesion in existing biological materials, researchers have developed a composite hydrogel that incorporates snail mucus, which prolongs the retention time of the hydrogel in the uterus and promotes endometrial regeneration. 48 Furthermore, the novel asymmetric Janus microgels simultaneously possess tissue fixation and anti-adhesion functions. 49 Stem cells, including embryonic, adult, and induced pluripotent stem cells, possess self-renewal and multilineage differentiation capacity. Stem cells are used in tissue repair and regenerative medicine applications because of their ability to differentiate into various cell types. 50 Animal studies have shown that mesenchymal stem-cell (MSC) therapy can improve endometrial regeneration and vascular reconstruction. When MSCs are used together with estrogen, they can alleviate endometrial fibrosis while promoting angiogenesis. 51 Bone marrow-derived MSCs regulate the expression of hormone receptors and participate in endometrial repair. 52 , 53 Both human amniotic membrane MSCs and human endometrial MSCs support endometrial regeneration. 54 , 55 Autologous CD133+ bone marrow stem-cell transplantation and autologous menstrual blood-derived stromal-cell transplantation can effectively treat severe IUAs by promoting endometrial regeneration and functional recovery. 56 , 57 Stem cells from multiple sources can reduce fibrosis, increase the number of glands, stimulate angiogenesis, and enhance endometrial thickness, ultimately improving pregnancy outcomes. 58 Stem-cell therapy has a promising future, but it is still in the early research stage at present. Limitations include low implantation efficiency, limited durability of therapeutic effects, and potential tumorigenic risk. 59 , 60 A new type of multifunctional microcapsule (A/G-Fe3O4-Se) has been developed for the prevention and treatment of IUAs. Microcapsules can be precisely distributed by magnetic field drive and can adapt well to the irregular uterine morphology. They have antioxidant and antibacterial effects and can promote endometrial regeneration. 61 A recently developed arrow-shaped micro-needle patch can prevent adhesion through a special arrow-shaped structure that is mechanically interlocked and stably positioned. Its tip promotes cell adhesion and tissue repair, whereas the base inhibits adhesion formation. The patch is soft in texture and can adapt to the complex structure of the uterine cavity. 37 Furthermore, the Janus nanofiber patch that employs advanced spinning and polymerization techniques has dual functions: one side offers anti-adhesion properties, whereas the other side exhibits anti-inflammatory and antioxidant effects. 62 A comparison of different anti-adhesion materials for the prevention of IUAs recurrence after TCRA is presented in Table 2 . Table 2 Comparison of Different Anti-Adhesion Materials for Prevention of Intrauterine Adhesion Reformation After TCRA Material Type Representative Method Advantages Disadvantages Suitable Population Physical barrier Balloon stent, IUD Simple, readily available Poor shape adaptation, higher infection risk Moderate-to-severe IUA Hormonal therapy Estrogen-eluting stent Local sustained release Limited long-term safety data Moderate IUA or above Hydrogel Self-healing hydrogel, Janus hydrogel Good adaptability to uterine cavity Rapid degradation Mild IUA Stem cells MSC, CD133+ cells High regenerative potential Safety not established, high cost Severe or recurrent IUA Microcapsule/patch Magnetically controlled microcapsule, micro-needle patch Precise delivery, degradable Limited clinical evidence Exploratory stage Abbreviations : IUD, intrauterine device; MSC, mesenchymal stem cells; TCRA, transcervical resection of adhesion. Comparison of Different Anti-Adhesion Materials for Prevention of Intrauterine Adhesion Reformation After TCRA Abbreviations : IUD, intrauterine device; MSC, mesenchymal stem cells; TCRA, transcervical resection of adhesion. The prognostic factors of reproductive outcomes after TCRA can be analyzed according to the time point of treatment. Preoperative factors include the severity of adhesions, the location of adhesions, the type of adhesions and previous surgical history. These factors can be used to formulate the initial treatment plan and provide preoperative consultation. Postoperative factors include the recovery of endometrial thickness, the normalization of the menstrual cycle, the results of secondary hysteroscopy, and the DEGO score. These factors can guide subsequent treatments, such as whether assisted reproduction is needed or adhesion release surgery should be performed again after the operation. Predictive models integrate multiple variables through statistics, machine learning or deep learning to predict reproductive outcomes and assess risks for individuals. Treatment outcomes and reproductive prognosis in patients with IUAs are closely associated with preoperative adhesion severity. As adhesion severity increases, treatment outcomes and prognosis deteriorate. 63 Adhesions are usually located in the center of the uterine cavity, on the side walls of the uterine cavity, or in both locations. IUAs are generally classified as mild, moderate, or severe according to adhesion extent and severity. 64 Adhesion severity is strongly associated with reproductive outcome. Recurrence is most frequent in severe adhesions. Moreover, owing to the poor development of the uterine endometrium, which cannot support fetal and placental growth, pregnancy complications may ensue. 2 , 16 In patients with IUAs, adhesions most commonly involve the uterine fundus, uterine body, isthmus, and cornual regions. Based on the morphology, texture, and degree of vascularization, adhesions can be classified into dense or membranous. Fundal and dense adhesions were most strongly associated with uterine-related infertility, whereas isthmic and filmy adhesions were less likely to cause infertility. 65 However, another retrospective study reported that initial adhesions located in the uterine cornua, isthmus, or extensively involving the uterine cavity were linked to increased risk of recurrence. 66 Studies indicate that increases in menstrual volume after hysteroscopic electrosurgery are smaller than those after adhesiolysis using cold scissors. 67 This difference may result from thermal damage to the endometrium caused by electrosurgery, leading to reduced menstrual volume. 68 A follow-up study of 70 patients compared the reproductive outcomes between scissors and electrosurgery groups. No significant intergroup differences were observed in the spontaneous abortion rates, term live birth rates, or average time to conception. 69 Similarly, a multicenter randomized controlled trial reported no notable differences between use of microscissors and electrosurgery in treatment efficacy, recurrence rate, pregnancy rate, or pregnancy-related complications, although use of microscissors was associated with increased postoperative menstrual volume. 70 A prospective study examining factors influencing adhesion reformation after TCRA found that pregnancy history, prior uterine surgery, preoperative endometrial thickness and volume, initial AFS score, presence of chronic endometritis, use of IUAs gel within 5 days postoperatively, and preoperative estrogen treatment may all affect postoperative adhesion recurrence and, consequently, reproductive outcomes. 4 Many patients with moderate-to-severe IUAs may require assisted reproductive technologies after surgery, particularly when natural conception is not achieved. Endometrial thickness ≥7 mm on the day of embryo transfer has been associated with improved pregnancy rates in some studies. 71–73 This threshold remains debated and lacks universal consensus. The extent of endometrial fibrosis remains a major determinant of reproductive outcomes, even after anatomical restoration of the uterine cavity. Therefore, restoring the function of the endometrium is just as important as restoring the anatomical structure of the uterine cavity. 2 The timing of embryo transfer affects pregnancy outcomes. A retrospective study suggested that a 90–180-day interval between TCRA and embryo transfer may be favorable. 74 However, prospective validation is needed before definitive recommendations can be made. Beyond fertility, IUAs also adversely impact obstetric outcomes after pregnancy is achieved. Munro et al reported an excess rate of adverse obstetric events in women with a history of IUAs, including preterm delivery, placenta accreta spectrum, placenta previa, peripartum hemorrhage, and hysterectomy, with some evidence suggesting that adjuvant therapies may mitigate these risks. 1 The main factors affecting pregnancy outcomes after TCRA, including disease-related factors, postoperative variables, and preventive measures, are summarized in Table 3 . Table 3 Summary of Factors Influencing Pregnancy Outcomes After TCRA in Patients with IUA Category Influencing Factor Effect on Pregnancy Outcome Disease-related Adhesion severity (mild/moderate/severe) Severe: significantly decreased pregnancy rate Disease-related Adhesion location (fundus/isthmus/cornua) Fundal and dense adhesions: poorer prognosis Postoperative Endometrial thickness (≥7 mm vs <7 mm) ≥7 mm: improved pregnancy rates reported (threshold debated) Postoperative Surgical technique (electrosurgery vs cold scissors) Cold scissors: better menstrual recovery, no significant difference in pregnancy rate Postoperative Time interval from surgery to embryo transfer (ET) 90–180 days: favorable in one retrospective study (unvalidated) Preventive measure Balloon stent vs IUD Balloon: more stable uterine microbiota Summary of Factors Influencing Pregnancy Outcomes After TCRA in Patients with IUA The predictive model estimates the probability of reproductive outcomes for individual patients by integrating multiple prognostic factors. The preoperative prediction model relies only on baseline data available before the initial surgery, such as the AFS score and Nasr classification. The postoperative prediction model incorporates variables obtained after TCRA, such as recovery of endometrial thickness, the results of secondary hysteroscopy, the DEGO score, and the return of menstruation. The following sections will introduce non-artificial intelligence and artificial intelligence-based prediction models, respectively. A retrospective study comparing the AFS, European Society of Gynecological Endoscopy, March, Nasr, and Chinese IUAs classification systems for predicting live birth rate after TCRA demonstrated that the Nasr classification showed the best predictive performance. 75 Decision tree prediction models can also assess the live birth rate after IUAs surgery. One of the key factors affecting model accuracy is the postoperative menstrual pattern. 76 Furthermore, another type of decision tree model can be used to predict IUAs recurrence. 77 Quantitative software-based analysis of the density of endometrial glandular orifices (DEGO) in hysteroscopy videos demonstrated significantly greater predictive ability for live birth, than the AFS score. These findings suggest that DEGO may serve as a novel objective indicator for predicting reproductive outcomes in patients with IUAs. 78 AI has advantages in analyzing large-scale medical images and electronic health record data and has become an important clinical decision-support tool. A study used the hysteroscopic images obtained after TCRA as training data and evaluated endometrial receptivity based on endometrial morphology and vascular distribution. The probability of conception within 1–2 years after surgery was predicted by combining four transfer learning architectures (ResNet, DenseNet, Inception, and EfficientNet) with a survival analysis model. Clinicians can formulate personalized treatment plans for patients based on the risk stratification established by the model. 79 An XGBoost model incorporating perioperative clinical parameters was developed to predict ongoing pregnancy following hysteroscopic adhesiolysis; the authors suggested it may aid in managing and categorizing patients with IUAs and determining the optimal approach to achieving pregnancy. 80 Multimodal learning has become an important area of research in medical AI. The generalization ability and clinical applicability of models can be enhanced through multimodal learning. Integration of multimodal data, including imaging features, clinical biomarkers, and electronic health records, has overcome the limitations of a single modality. 81 , 82 A multimodal model has been developed to predict postoperative reproductive outcomes in patients with IUAs. This model utilizes electronic health records and hysteroscopy images to assist clinicians in providing personalized guidance to patients after TCRA surgery. 83 AI imaging tools should be rigorously validated before being applied in clinical practice. Mlodawski et al 84 demonstrated that an AI-based fetal biometric tool (SonoCNS) showed excellent repeatability for certain parameters (BPD, HC; ICC > 0.75), but had poor reliability for other parameters (CM, Vp; ICC < 0.75). This indicates that the performance of artificial intelligence is dependent on the specific application and is not universally reliable. At present, reproductive outcome prediction models used for intrauterine adhesions (IUAs) have not been systematically validated for repeatability of measurements across different adhesion subtypes, operators, or devices. However, for morphometric data generated using AI, there is a risk that these measurements cannot be reproduced without such verification. Future AI research on IUAs should report ICC values for intraobserver and interobserver reliability and compare them with results manually annotated by experts. The temporal evolution of predictive models for clinical pregnancy outcomes after TCRA is summarized in Figure 2 . Direct cross-study comparisons are not feasible because of variations in outcome definitions, cohort characteristics, and performance metrics. Figure 2 Temporal evolution of predictive models for pregnancy outcomes after transcervical resection of adhesion (TCRA) in patients with intrauterine adhesions (IUAs). Bubble sizes are arbitrarily scaled for visual clarity and do not imply a quantitative comparison of predictive performance across different models because of heterogeneity in study designs, evaluation metrics, and validation approaches. Data sources for each model are listed below the bubbles. Timeline of predictive models for pregnancy outcomes after TCRA from 1988 to future AI-assisted surgery. Temporal evolution of predictive models for pregnancy outcomes after transcervical resection of adhesion (TCRA) in patients with intrauterine adhesions (IUAs). Bubble sizes are arbitrarily scaled for visual clarity and do not imply a quantitative comparison of predictive performance across different models because of heterogeneity in study designs, evaluation metrics, and validation approaches. Data sources for each model are listed below the bubbles. AI though some progress has been made in predicting reproductive outcomes in patients with intrauterine adhesions (IUAs), many limitations remain. Most prediction models have small sample sizes and lack external validation. Model calibration has not yet been systematically reported. In addition, the lack of standardized definitions of key variables, such as the threshold for endometrial thickness and the severity of adhesions, across studies has affected the reproducibility of the research results. Therefore, large-scale, multicenter, prospective clinical studies with standardized research protocols are urgently needed to develop robust and clinically useful diagnostic and therapeutic tools. TCRA has become the gold standard for treating IUAs. Hysteroscopy enables direct visualization of intrauterine lesions, allowing precise adhesion removal while maximizing preservation of healthy endometrium. TCRA is the standard surgical method to restore menstruation and enhance fertility in patients with IUAs. Therefore, it is widely used in clinical settings. 85 Algorithms such as deep neural networks require large-scale datasets for training. These trained models can subsequently generate predictions using previously unseen data. These algorithms are widely used for image and video recognition and classification. In the field of medical imaging, deep learning is widely used for diagnosing cancer and detecting polyps. However, real-time surgical guidance systems remain technically complex and have not yet achieved widespread clinical implementation. 86 Many surgical complications arise from the visual limitations of the surgeon, and deviations caused by such limitations may affect the patient’s postoperative recovery. 87 AI-assisted surgical systems are expected to reduce judgment errors caused by human visual limitations and improve surgical outcomes. However, an AI-assisted surgical guidance system specifically designed for IUAs has yet been developed. In contrast, intraoperative optical imaging and ultrasound have been reported to support real-time anatomical localization during endometriosis surgery. Augmented/virtual reality and artificial intelligence-assisted recognition are still in their early stages. These findings suggest that similar navigation techniques may have potential applications in IUA surgery. 88 The various factors influencing the pregnancy outcomes of IUAs patients and future optimization directions are depicted in Figure 3 . Figure 3 Comprehensive management strategies for patients with intrauterine adhesions (IUAs) and optimization pathways for pregnancy outcomes. The diagram presents strategies for managing intrauterine adhesions. It includes four sections: Challenges and Foundations, Postoperative Precise Assessment and Management, Intraoperative Intervention Strategies and Optimize Orientation and Future Outlook. Challenges and Foundations list the severity of adhesion and previous surgical history. Postoperative Precise Assessment and Management includes regular follow-up with hysteroscopy, analysis of endometrial receptivity and individualized assisted reproduction. Intraoperative Intervention Strategies feature anti-sticking material, uterine cavity support and estrogen therapy. Optimize Orientation and Future Outlook suggests a prospective multicenter study, establishing a predictive model and multidisciplinary collaborative diagnosis and treatment. Diagram of strategies for managing intrauterine adhesions, including challenges, interventions and future outlook. Comprehensive management strategies for patients with intrauterine adhesions (IUAs) and optimization pathways for pregnancy outcomes.

Multiple factors influence treatment efficacy and pregnancy outcomes in patients with IUAs. The restoration of the anatomical structure of the uterine cavity is a key factor influencing reproductive prognosis. IUAs severity and baseline characteristics, such as prior intrauterine surgery, also influence pregnancy outcomes. Postoperative management measures, such as regular hysteroscopic monitoring, and individualized assisted reproductive protocols, also influence pregnancy success rates. The use of anti-adhesion materials, placement of intrauterine balloons, and estrogen therapy can reduce the risk of re-adhesion and promote endometrial repair. In addition, postoperative hysteroscopy follow-up, endometrial receptivity assessment, and biomarker analysis are important components of assisted reproduction after TCRA. Prediction of postoperative reproductive outcomes may help clinicians identify high-risk patients and optimize treatment strategies to improve live birth rate. It is essential to conduct forward-looking, randomized, multicenter clinical studies, systematically investigate the effects of different postoperative intervention measures on reproductive outcomes, and establish reliable prediction models to optimize treatment plans. Integration of imaging data, molecular biomarkers, and clinical scoring systems may improve prediction of adhesion recurrence and pregnancy outcomes. There remain limitations in screening, treatment, and follow-up in the current IUAs management system. It is necessary to promote a transition toward a multidisciplinary diagnostic and treatment model. Current AI-based prediction models should remain investigational until externally validated in sufficiently large, independent, and preferably multicenter cohorts.

Pregnancy refers to the presence of a gestational sac confirmed by ultrasound. Clinical pregnancy is defined as pregnancy confirmed by ultrasound visualization of an intrauterine gestational sac with fetal cardiac activity. Ongoing pregnancy denotes a pregnancy progressing beyond 12 weeks of gestation. Live birth indicates delivery of a viable infant after ≥28 weeks of gestation. Natural conception specifically refers to pregnancy achieved without assisted reproductive technology (ART). Infertility is defined as the failure to achieve pregnancy after 12 months of regular unprotected intercourse or after appropriate ART. Adhesion recurrence refers to the reformation of intrauterine adhesions confirmed by hysteroscopic or imaging evaluation following initial surgical treatment. These definitions are applied consistently throughout this review.