Revolutionizing Women’s Health: Artificial Intelligence’s Impact on Obstetrics and Gynecology

INTRODUCTION

Artificial intelligence (AI) is the term for a digital computer system designed to simulate how the human brain processes information cognitively. Artificial Intelligence is referred to as neural connections because it is organized into multiple neural nodes, much like the organization of nerves in the brain.^1,2 Artificial neural networks (ANNs) have advanced as a result of AI, and this includes a dependable logical structure that can analyze complicated data^3–5 The neurons establish many synaptic connections to exchange data, enabling them to determine the most likely solution collectively. Establishing these many links allows computers to emulate cognitive processes, including logical thinking, to choose the most likely solution. Sophisticated algorithmic AI software is being used in the field of medicine to evaluate vast quantities of data, therefore aiding in illness avoidance, diagnosis, and patient monitoring.² Artificial intelligence technology is used in gynecological surgery to enhance surgeons’ ability to identify crucial anatomical features, decrease complications, reduce the duration of the operation, and provide a realistic teaching environment for aspiring surgeons. Utilizing multifaceted [three-dimensional (3D)] printers using AI enables the production of materials that closely resemble actual tissues, facilitating realistic training for trainees. Moreover, 3D imaging provides superior depth awareness compared to its multifaceted [two-dimensional (2D)] equivalent, enabling the surgeon to develop surgical plans based on tissue depth and dimensions. Despite its limits, AI holds the potential to enhance patient prediction and administration, lower healthcare expenses, and enable obstetrics and gynecology (OB/GYN) professionals to streamline their burden and enhance productivity and precision by integrating AI technologies into existing everyday practices. Artificial intelligence has tremendous capacity to assist professionals in making informed decisions, arriving at accurate diagnoses, and enhancing case management. It can lower medical expenses by diminishing healthcare mistakes by offering more reliable forecasts. Artificial intelligence technologies can effectively offer precise knowledge to a wide range of individuals in healthcare environments, while the availability of increasingly comprehensive information is necessary. Artificial intelligence in healthcare systems has seen a significant surge in worldwide expansion in recent years. Based on statistical data, the projected trend indicates a rise in investment in AI from $2.1 billion to $36.1 billion by 2025.³ Successful use of computer intelligence in medicine necessitates education and coordination among stakeholders. An example of such technology is IBM’s Watson for oncology, a renowned AI system that has garnered user repute. Its primary function is to collect prescription data from several medication libraries and scientific publications.⁴ Artificial intelligence may assist professionals in decision-making and enable physicians to generate better, confident choices. Nevertheless, it is crucial to bear in mind that it must maintain the value of practical expertise.³ Healthcare mistakes rank as the third most common source of mortality in the United States (US), and AI has the potential to aid in their reduction by enhancing accuracy in interpretation and reducing the burden that might result in the omission of information, and improving the efficiency of their performance reducing the mortality.⁵ This review presents several classifications of AI, explores their possible programs, examines related studies, discusses the advantages and disadvantages, and speculates on the development of AI. Additionally, it explores AI’s potential benefits in OB/GYN. In gynecology, the integration of machine learning, particularly deep learning (DL) techniques, has revolutionized medical analysis and diagnosis. Supervised learning, encompassing regression and classification, aids in predicting outcomes such as gestational age or identifying anomalies through labeled datasets. Regression models predict continuous variables such as fetal growth, while classification models diagnose conditions such as preeclampsia. Unsupervised learning methods such as clustering identifiable patterns within unlabeled data, and distinguishing patient groups based on symptoms or genetic traits. Dimensionality reduction techniques further refine complex gynecological data, extracting critical features for improved analysis Figure 1. Through these machine learning paradigms, gynecologists gain invaluable insights for early detection, personalized treatment, and better maternal and fetal care.

Tracking the fetal heart rate (FHR) enables medical professionals to monitor the developing fetus and identify high-risk issues. Additionally, it provides a comprehensive analysis of the basal FHR, its variability, accelerations, decelerations, uterine spasm strength, and variations in FHR patterns.⁶ Artificial intelligence is now utilized to observe the FHR during childbirth by evaluating cardiotocographs (CTGs) and predicting potential results. According to an early study, incorporating AI into obstetrics, particularly antepartum surveillance, might have significant advantages.² This method would mitigate the disparities among obstetricians in evaluating intrapartum tracking, yielding a more dependable and reproducible outcome for every study and eventually diminishing neonatal and maternal problems and mortality. The AI algorithms may also deliver corroborating evidence in situations where unexpected negative results may lead to legal action. In 2018, a study called computer-aided fetal evaluator (CAFE) examined the potential of using AI to evaluate CTG data, which is used in analyzing FHR patterns. The findings indicated that the AI system exhibited comparable content comprehension to domain experts and could identify mistakes. One of the drawbacks of this research was the presence of discrepancies among specialists in evaluating certain data, particularly about variances. This hindered the capacity to determine whether the mistake in interpretation was due to a flaw in the system or in reaching a consensus among professionals in the area.⁷ These presented a challenge in demonstrating the system’s optimality and verifying its accuracy. Further comprehensive study is necessary to improve the adaptability of information analysis. The CTG technique emerged half a century ago, and the frequent discrepancies among experts need a method that reduces errors and standardizes interpretation during CTG analysis. The INFANT research protocol assesses the capacity of AI over CTG during labor to aid practitioners in making personalized decisions on the most effective management approach.⁸ The objective is to enhance the dependability of FHR readings, facilitate the clinician in understanding and formulating decisions, and reduce the workload by improving efficiency. Perinatal asphyxia is a substantial global issue and developing an effective method to monitor FHR will enhance medical attention and reduce adverse consequences. Therefore, an alarm monitor that alerts the clinician to probable fetal discomfort may prompt quick action and facilitate childbirth treatments as needed. Additionally, it may provide comfort and prevent unneeded intervention in mothers who do not exhibit signs of fetal stress. The System 8000 is a computer system designed to evaluate the prenatal FHR within the range. The system analyzes fluctuations in FHR and identifies the magnitudes linked to low oxygen levels by identifying decreases and alterations in unpredictability.⁹ A reduction in variability is the greatest reliable indicator of fetal deterioration. However, there is regrettably considerable variability among observers when interpreting this data. The use of AI has the potential to reduce the disparity among gynecologists, resulting in a better dependable understanding of variance, less need for excessive interventions, and accelerated delivery if required.⁹ Novel methods should be developed to deal with the problem of newborns and maternal mortality. The authors thought that AI might be used in outpatient treatment by creating home monitoring capable of appropriately monitoring individuals who pose a danger. When combined with telemedicine, this technology may help identify any issues sooner and provide patients and clinicians confidence that a secure surveillance system is in place, even when the patient is receiving treatment at a clinic or hospital that is not affiliated with them. Additionally, it offers a warning system that alerts users to potentially harmful FHR values and instructs them to contact their doctor so that the in-home AI analysis can determine the best course of action. The neural network-based recognition methods this research utilizes can resolve interpretive differences across experts. It also noted how easy it is to obtain outpatient monitoring since most houses have cell phones and Internet connections. The AI integration with Doppler ultrasonography has shown to be a financially viable solution. To prevent the physician from being unnecessarily alarmed about potential fetal risk, the equipment was also able to filter out measurements with unclear meaning, such as faux accelerations and pseudo decelerations brought on by strong fetal movement. The study’s drawback was that the suggested approach was only tried on one participant, and even though it worked, further research is required to confirm the findings and verify this intervention. Home-based pregnancy monitoring, coupled with the ability to guide decision-making and treatment via telecommunication, would potentially reduce maternal and newborn mortality and have an important function in the diagnosis of pregnancy problems.^8,9 Artificial intelligence integrates with a remote health survey system via the internet, enhancing ultrasound detectors efficiency. Artificial intelligence analyzes FHRs, nonlinear parameters, and the fetal wake–sleep cycle in real time. This synergy streamlines data collection, optimizing prenatal care remotely. Utilizing AI algorithms, the system interprets ultrasound data, tracking FHR variations and nonlinear parameters for comprehensive assessments. Moreover, it precisely monitors the fetal wake–sleep cycle, aiding in developmental evaluations.⁹ This interconnected network empowers remote health systems, offering precise, timely insights into fetal well-being, enabling proactive interventions and personalized care, and ultimately advancing prenatal health care through innovative technological collaborations (Fig. 2).

PREGNANCY-INDUCED DIABETES

Present-day gestational diabetes mellitus (GDM) monitoring is expensive and burdensome for expectant mothers. Artificial intelligence can help in the process of identifying alternatives to the present norm. The most recent screening guidelines, according to the United States Preventive Services Task Force, require further diagnosis assessment and glucose oral tolerance examination to be performed after 24 weeks of pregnancy. They discovered that there was not enough data to screen before 24 weeks of pregnancy.¹⁰ Screening for GDM is crucial to avoid issues including neonatal hypoglycemia, big for gestational-age babies, and a rise in cesarean deliveries. In addition to a rise in cesarean births, GDM also has more expensive effects; however, these might be avoided with the availability of a more practical screening tool. An earlier study recommended utilizing a web-based diagnostic instrument that clinicians and patients could utilize jointly to see if employing an AI algorithm to test for GDM could be less expensive and more accessible for patients than the current guidelines. An ANN that was converted into a Java applet and made available online served as the foundation for the calculator.¹¹ The AI system was developed using a model known as multilayer perceptual neurons, capable of transmitting and receiving data, thus the term algorithm-back propagation (BP). The calculator considers risk factors, including cigarette smoking, reduced-fat diet, pounds, high blood pressure, hyperlipidemia, and ethnicity. The present ANN model on the website develops and learns as it becomes subject to additional instances the authors arrived at without the ultimate goal of ultimately helping to reduce health expenditures, even if AI is less effective than the conventional examination.¹² Its ease of use, which makes it useful for everyday usage, is a bonus.

PREMATURE LABOR

To forecast prenatal outcomes in women who were asymptomatic and had a short cervix length, a study combined AI with amniotic fluid (AF) proteomics and metabolomics, either in conjunction or separately with images, demographics, and medical information.¹³ They utilized a kind of AI known as DL. Since it contains more ANNs than other subtypes of AI, this type is more suited for multiomics-based biological system investigations and can function with larger amounts of data. The main risk factor for premature at the moment is a short cervix length, yet numerous mothers having this issue deliver their babies to term. Amniocentesis is currently utilized at numerous facilities to assess additional hazards for these women, including infections and inflammation. The participants’ AF was also investigated for the omics, including metabolomics, to identify novel markers that may be connected to premature delivery. This may increase the prediction relevance and reliability for women who are at risk of unfavorable outcomes. Furthermore, compared to existing risk indicators with the value of short cervical length and history of premature delivery, it may assist doctors in effectively stratifying individuals who are at risk of PRT birth.¹³ Depending on the circumstances, clinicians may use this tool to guide their treatment and decide if to provide prenatal steroids and/or cervical cerclage, or they can use it to just monitor the patient. The sample quantity represented one of the research’s limitations. To guide treatment options for such patients, the investigation found that DL provided a more accurate predictor of the perinatal outcome in asymptomatic women with short cervix length. It also indicated that more research is necessary to investigate the correlation between AF omics and premature cervix decrease. To reduce costs associated with incorrectly diagnosing PRT labor and avoid needless hospital stays and processes, a study highlighted the significance of utilizing AI technology.¹⁴ In the interim, individuals who are in true labor should receive treatment more quickly to avoid potentially dangerous outcomes for both the mother and the unborn child. The electrohysterography (EHG) impulses were classified using three distinct machine learning algorithms, allowing the researchers to distinguish between genuine labor and properly diagnose premature labor. The random algorithm worked the best since it could process more data, was accurate, and displayed a strong learning capacity. Consequently, the algorithm was capable of forecasting premature birth with a higher precision.

LABOR AND DELIVERY

If labor starts earlier than or later than the advised window, it can give rise to complications such as preterm (PRT) or postterm (POT) pregnancy. To gain a greater awareness of the biological mechanisms governing labor, genetic microarray sequencing of the myometrial events during pregnancy was explored.¹⁵ A gene schematic utilizing improvements in AI technology, was developed which helped identify relevant information on myometrial activation from a large amount of data. However, human research has yet to be done to support this. Artificial intelligence was employed to assist with parturition-related problems and suggested more research to comprehend the genes involved in human parturition and corroborate these findings.¹⁶ Other factors, such as challenges during pregnancy, must also be considered because of their potential impact on the development of myometrial genes. The genes engaged during labor can result in effective medical interventions when necessary, helping alleviate parturition conditions such as POT pregnancy and PRT delivery and lowering the perinatal morbidity and mortality associated with these issues. The MetaCore program is an excellent example of innovative technology that might help with the difficult task of determining if the expression of genes is regulated in the myometrium.¹⁷ It is composed of data-evaluation resources, gene lists, and an information repository. Its disadvantage is that it needs a predetermined method or prior knowledge of the problem. Numerous factors may contribute to PRT or POT labor, some immediately obvious or acknowledged.^15–17 Other systems, such as neural networks, may overcome these drawbacks since they do not need prior knowledge or adhere to a set protocol. With MetaCore, many data, including genes and related factors, can be nonlinearly examined. One disadvantage is that most computer projections have models and must be tested on humans to ensure they are viable and accurate. It also requires much analysis and research.

ASSISTED REPRODUCTIVE TECHNOLOGY (ART) (IN VITRO FERTILIZATION)

Assisted reproductive technology encompasses various fertility treatments, including in vitro fertilization (IVF). Also, IVF, a cornerstone of ART, involves fertilizing eggs with sperm outside the body. This method aids conception for those facing infertility, employing controlled laboratory conditions for embryo development before transferring viable embryos to the uterus for potential pregnancy. Information extraction and AI techniques were utilized to develop a computer program capable of accurately forecasting pregnancy outcomes in the context of IVF.¹⁸ Data mining (DM) utilizes AI and sophisticated statistical techniques to uncover relationships inside extensive datasets. Data mining not only retrieves relevant knowledge but also identifies other crucial factors that may impact the result, hence expanding the pool of usable data.¹² Identifying emerging patterns that impact the likelihood of effectiveness in IVF is crucial for patients and clinicians to establish reasonable expectations for the outcomes. To assist physicians in forecasting pregnancy success rates, a hybrid intelligent model was developed. This system used DM to combine genetic algorithm-based and decision-tree learning approaches, which were employed to retrieve knowledge from the records of IVF patients. Researchers discovered that this approach aided in forecasting results and proposed customized IVF therapy based on specific patient attributes.¹⁹ An inherent limitation of the research is that the model they constructed relied solely on information from a single IVF institution. Nevertheless, if the centers collaborated to exchange knowledge, their combined data may substantially broaden the representation of a larger community with enhanced precision. Additional research indicates that using ANN systems may be effective in forecasting IVF results. This can be achieved using a training vector quantizer that enables generalization and utilizes standard variables to boost the prediction capability.¹⁸ Both IVF and AI also have the potential to select the most viable oocytes and embryos. An AI system was proposed integrating AI techniques to extract texture descriptors from a picture, namely the regional binary structure, and then assemble them using an ANN.²⁰ These findings demonstrated superior performance in comparison to existing techniques and have the potential to facilitate the unobtrusive and accurate selection of optimal oocytes or embryos. Moreover, researchers emphasize the benefits of this technique in choosing the most viable embryos, particularly in nations whose laws prohibit the choice of embryos based on sex.

ONCOLOGIC SCREENING

Neural network models are now utilized for providing prognoses in patients who have ovarian cancer. Ovarian cancer is a term used to describe a diverse group of tumors that vary in their characteristics, such as their appearance under a microscope and how they appear in patients, including the stage of the tumor. Artificial neural network achieved remarkable efficiency in predicting survival.²¹ The AI algorithms they created can provide a precise diagnosis. An AI program was developed that could forecast the likely course of ovarian cancer in women more accurately than existing techniques.²¹ Additionally, it can anticipate the most optimal course of therapy based on each patient’s individual diagnosis. The long-term prognosis for advanced ovarian cancer is unfavorable; thus, there is a need for more precise and focused treatment options. At present, there is no available method for detecting ovarian cancer despite its prevalence as a prevalent form of gynecological cancer. Consequently, the majority of cases are detected in later stages, resulting in a significant 5-year death rate. The AI neural net effectively tracked the intricate interplay of micro-RNA. It successfully detected nearly 100% of ovarian cancer anomalies, in contrast to an ultrasound screening test that only found aberrant findings fewer than 5% of the time. This simple diagnostic involves the measurement of micro-RNAs from a blood sample. It may be crucial for the future treatment of ovarian cancer. A novel AI approach had been designed to examine ovarian cancer cells via biomarkers and detect irregularly shaped nuclei associated with highly malignant tumors. This system is designed to identify individuals at risk of developing such tumors.¹⁷ An AI system may be used during regular biopsies to scan for risk factors associated with DNA instability and to choose appropriate therapeutics based on the findings. The presence of malformed nuclei in aggressive ovarian cells has shown the ability of the immune systems to be evaded. This suggests that immune-targeted therapies, such as once-immunotherapy, may elicit a reaction. Artificial intelligence has been integrated into the field of cancer via commercialized programs that use algorithms to pair patient information with ongoing clinical trials around the country and corresponding experimental medications for each person;^22–24 IBM’s Watson for Oncology utilizes AI in combination with medical information to provide guidance in cancer care, demonstrating notable efficacy in the case of breast cancer patients.²⁵ Moreover, AI has surpassed human specialists in the interpretation of cervical precancer pictures. The current screening procedure involves visually examining the material obtained from a Papanicolaou (PAP) smear and using acetic acid to enhance the visibility of whiteness in the tissue, which may indicate the presence of illness. Although it is convenient and inexpensive, it needs more precision. The AI deep training algorithms have the capability to collect a vast amount of photos pertaining to cervical cancer screening and accurately detect abnormal cells. Additionally, it guarantees that the test is easy, precise, and economical, in contrast to the existing approach. Only a small amount of education is necessary, and the outcomes are instant; thus, patients may get therapy during their appointments.^22–24

GYNECOLOGICAL SURGERY PROCEDURE

Physical AI has been more extensively used in surgical procedures than virtual AI. The digital AI system utilizes known individual parameters, repeated patterns, and treatment algorithms to forecast the result, unlike the surgical domain, which involves several independent variables. The parameters that contribute to the challenge of creating an algorithm include the varying uniformity of distinct tissues, the proficiency of individual surgeons, the modifications made during surgery, and the differences between patients and their specific conditions.²⁵ Artificial intelligence has supported gynecological treatment in areas such as imaging and spatial awareness. Artificial intelligence can enhance the surgeon’s capabilities by offering superior images before and during surgical procedures. The development of three-dimensional printing (3DP) technology, which accurately reproduces the surgery site, surpasses its 2D equivalent in terms of precision, offering a more faithful representation of the real model.^26,27 This facilitates a more precise preoperative strategy, consequently reducing mistakes in the operating theater. Three-dimensional printing may offer a variety of materials that closely mirror the tissues faced, allowing trainees to practice realistically and enabling advanced preoperative planning.²⁷ In essence, a 3DP picture assists in guiding the medical procedure and enhancing understanding of the specific region, therefore safeguarding the adjacent structures. It was beneficial in a single case study with profound infiltrated endometriosis (DIE).^26,27 A 3D-printed model was created using preoperative magnetic resonance imaging (MRI) data and then compared to the surgical results in a retrospective analysis. The surgeons agreed that the 3D image helped with the anticipated procedure since it made it possible to determine the depth, width, and involvement of the surrounding structures, which helped to direct the next operations. The aforementioned models have an opportunity to serve as additional tools for preoperative planning when it is challenging to analyze length and spatial relationships using just standard 2D photographs. Artificial intelligence has thus contributed to reducing the duration and improving the precision of surgical procedures, reducing surgical problems. This has been achieved via the use of augmented reality. Augment realism is using technology to digitally recreate real-world items and improve them to generate a more informative visual representation.^28,29 A critical assessment of augmented reality in surgery highlights many limitations of this technology, including heightened expenses, apprehension over system latency, and the cumbersome and impracticable nature of the head-mounted display during lengthy surgical procedures.³⁰ Moreover, augmented reality has the potential to induce vomiting and dizziness as a result of “simulator sickness,” which could be very undesirable in the context of surgical procedures. Augmented realism may provide excessive information, leading to visual overload and diverting the surgeon’s attention from the primary objective. It might be best beneficial in a stationary environment since movement tissues such as the uterus are not well-suited for this technique. Notwithstanding several drawbacks, the enhanced realism had generally advantageous effects, including enhanced accuracy, security, and cost efficiency in carrying out processes.^29,30 However, more research and enhancements are necessary to exploit the potential advantages. Additional instances of computer systems aiding in surgical procedures may be seen in robotics and computer-assisted platforms. These systems provide the option of choosing a minimally invasive approach throughout surgery, which may effectively decrease the need for intrusive operations, mitigate their associated risks, and shorten operating durations. Certain robots can reduce vibrations, hence enhancing precision.²⁵ Moreover, the use of advanced tools, such as 3D laparoscopic surgery, enhances the ability to see within the body and offers the capacity to improve surgical results.³⁰ Artificial intelligence has improved spatial awareness by alerting surgeons to hidden important arteries, promptly recognizing them, and safeguarding crucial structures. One instance is the separation of the ureters during gynecological surgery. An investigation was conducted using an AI-based endoscope system that used algorithms to assess the thickness and positioning of the ureters (Fig. 3). This research shows that the utilization of such technology significantly improved both accuracy and safety.³¹

SUMMARY

The use of machine intelligence in health care may provide substantial improvements, yet it is crucial to take into account the potential drawbacks of AI in machines. The resolution of moral concerns, such as the possibility of individual prejudices in the development of software computations, requires attention.³³ Healthcare projections may differ based on characteristics such as ethnicity, inheritance, and sex, amongst additional variances. Refrain from including such elements to avoid overestimating or underestimating patient risk factors. For AI analytics in health care, clinicians would be responsible for ensuring the proper development and use of AI algorithms.³⁴ Health care must adhere to ethical principles to maintain trustworthiness, while health care should emphasize the patient’s well-being. The potential of AI remains intriguing since it can lower healthcare expenses and alleviate the burden on clinicians by assisting in decision-making processes. Artificial intelligence could increasingly integrate into clinical practice. Machine learning tools, such as the IDX-DR AI instrument developed by IDx Technologies Inc., Coralville, IA, USA can now understand their leads without needing a clinician’s review. This device specifically detects mild diabetic retinopathy.³⁵ Given its independence from expert interpretation, this software may be used by a greater number of primary care doctors, possibly alleviating the burden on specialists, and enabling interpreting by professionals of all sorts. This can be employed in the field of obstetrics for the interpretation of FHR and CTG readings. As shown by specific cases, AI-based machine learning design [computer-aided design (CAD)] has sometimes reduced accuracy in interpreting mammograms.³⁶ The performance of AI in diagnosing breast cancer was demonstrated to surpass that of a conventional specialist. Nevertheless, in actual implementation, a single radiologist examines these pictures and may disregard CAD recommendations as a result of prejudice.³⁴ However, if two experts examine the picture and have conflicting opinions, it is quite probable that the image will be sent for further examination. Therefore, the cost-effectiveness of AI in comparison to interpreting by two radiologists remained uncertain in this investigation. Another drawback is that AI-powered CAD needs to have the ability to provide a rationale for its decision-making process. Therefore, when there is a misunderstanding by the software, it becomes easier to establish whether the responsibility lies with the manufacturer or the radiologist who reads the data. It is necessary to create organizations capable of developing standards that can verify and assure the quality and correctness of products. Furthermore, the algorithms must perform well in various situations that accurately simulate the practical circumstances in which algorithms are used in clinical applications.³⁷ Simulating real-world scenarios becomes more feasible using substantial data, which may be acquired via obtaining medical information. Nevertheless, maintaining patient anonymity is a difficulty when accessing personal information. Blockchain solutions can ensure the confidentiality of medical records. That could enable concurrent data exchange across centers, integrate it into the AI software, and facilitate its continuous expansion of records, resulting in enhanced accuracy.³⁷ Medical professionals must comprehensively understand AI to modify it effectively, enabling the machine to provide precise information. Moreover, the machine needs adaptability in assimilating new knowledge, necessitating a continuous learning process and subsequent adjustment. Furthermore, the data should accurately reflect the characteristics of the population under assessment in a practical clinical environment. Although AI faces difficulties, it can revolutionize individuals’ concerns by offering precise diagnoses, reducing the workload for clinicians, lowering healthcare expenses, and establishing a standard analysis in cases where specialists have varying interpretations. Advancements in medical AI will persist. Healthcare professionals should wholeheartedly adopt them in their final days while also exercising caution and, if needed, acknowledging the benefits and limitations to consistently provide optimal patient care.

CONCLUSION

Researchers have examined various manifestations of AI, previous studies on AI, its benefits and drawbacks, possible obstacles, and the prospective use of AI in OB/GYN. This thorough analysis demonstrates that AI has great potential in addressing diagnosis obstacles and enhancing therapeutic methods and individual results in OB/GYN. Additional research is required to mitigate biases in algorithm development and enhance the system’s flexibility, allowing for the integration of emerging clinical information with emerging technological advancements. Professionals should additionally implement precautions to guarantee the validity and accuracy of the analysis. Artificial intelligence is intended to supplement professionals, instead of replacing them by assisting in the process of decision. From an ethical standpoint, using medical information may compromise personal anonymity due to the extensive information needed for AI algorithms to access comprehensive and diverse demographic statistics seen in healthcare environments. This allows for more realistic and precise predictions.

ORCID

Akila K https://orcid.org/0000-0002-7066-8239

Sovan Bagchi https://orcid.org/0000-0003-3507-1944

Vinaya Vijayan https://orcid.org/0000-0003-1182-9474

Polevoy Georgiy https://orcid.org/0000-0002-3300-3908

Sasikala M https://orcid.org/0000-0003-4298-133X

Devika S Kumar https://orcid.org/0000-0002-4442-9197

REFERENCES

1. Mhaskar HN, Zahavy T, Liao Q, et al. Learning functions: When is deep better than shallow. Adv Neural Inf Process 2017;30:1542–1552. DOI: 10.48550/arXiv.1603.00988.

2. Malani SN IV, Shrivastava D, Raka MS. A comprehensive review of the role of artificial intelligence in obstetrics and gynecology. Cureus 2023;15(2):e34891. DOI: 10.7759/cureus.34891.

3. Schwendicke F, Chaurasia A, Wiegand T, et al. Artificial intelligence for oral and dental healthcare: Core education curriculum. J Dent 2023;128:104363. DOI: 10.1016/j.jdent.2022.104363.

4. Zhao X, Zhang Y, Ma X, et al. Concordance between treatment recommendations provided by IBM Watson for oncology and a multidisciplinary tumor board for breast cancer in China. Japanese J Clin Oncol 2020;50(8):852–858. DOI: 10.1093/jjco/hyaa051.

5. Singh H, Graber ML. Improving diagnosis in health care—The next imperative for patient safety. New Engl J Med 2015;373(26):2493–2495. DOI: 10.1056/NEJMp1512241.

6. Khalil M, Ganapathy R, Mahsud–Dornan S. et al. Computer-assisted fetal monitoring: Current insights and future directions in obstetrics. Int J Womens Health 2021;13:1227–1239.

7. Guijarro–Berdiñas B, Alonso–Betanzos A, Fontenla–Romero O. Intelligent analysis and pattern recognition in cardiotocographic signals using a tightly coupled hybrid system. Artificial Intelligence, 2002;136(1):1–27. DOI: 10.1016/S0004-3702(01)00163-1.

8. Brocklehurst P, INFANT Collaborative Group. A study of an intelligent system to support decision making in the management of labour using the cardiotocograph–the INFANT study protocol. BMC Pregnancy Childbirth 2016;16:1–5. DOI: 10.1186/s12884-015-0780-0.

9. Dawes GS, Moulden M, Redman CW. System 8000: Computerized antenatal FHR analysis 1991;19(1–2):47–51. DOI: 10.1515/jpme.1991.19.1-2.47.

10. Shen J, Chen J, Zheng Z, et al. An innovative artificial intelligence–based app for the diagnosis of gestational diabetes mellitus (GDM-AI): Development study. J Med Internet Res 2020;22(9):e21573. DOI: 10.2196/21573.

11. Shankaracharya, Odedra D, Mallick M, et al. Java-based diabetes type 2 prediction tool for better diagnosis. Diabetes Technol Ther 2012;14(3):251–256. DOI: 10.1089/dia.2011.0202.

12. Loku L, Fetaji B, Fetaji M. Prevention of diabetes by devising a prediction analytics model. In: 2020 International Congress on Human-Computer Interaction, Optimization and Robotic Applications (HORA); 2020 Jun 26, IEEE.

13. Bahado–Singh RO, Sonek J, McKenna D, et al. Artificial intelligence and amniotic fluid multiomics: Prediction of perinatal outcome in asymptomatic women with short cervix. Ultrasound in Obstet Gynecol 2019;54(1):110–118. DOI: 10.1002/uog.20168.

14. Lee KS, Ahn KH. Application of artificial intelligence in early diagnosis of spontaneous preterm labor and birth. Diagnostics (Basel) 2020;10(9):733. DOI: 10.3390/diagnostics10090733.

15. Salomonis N, Cotte N, Zambon AC, et al. Identifying genetic networks underlying myometrial transition to labor. Genome Biol 2005;6(2):R12. DOI: 10.1186/gb-2005-6-2-r12.

16. Schuler G, Fürbass R, Klisch K. Placental contribution to the endocrinology of gestation and parturition. Animal Reprod 2018;15(Suppl. 1):822–842. DOI: 10.21451/1984-3143-AR2018-0015.

17. Nikolsky Y, Ekins S, Nikolskaya T, et al. A novel method for generation of signature networks as biomarkers from complex high throughput data. Toxicol Lett 2005;158(1):20–29. DOI: 10.1016/j.toxlet.2005.02.004.

18. Letterie G, Mac Donald A. Artificial intelligence in invitro fertilization: A computer decision support system for day-to-day management of ovarian stimulation during in vitro fertilization. Fertil Steril 2020;114(5):1026–1031. DOI: 10.1016/j.fertnstert.2020.06.006.

19. Jurisica I, Mylopoulos J, Glasgow J, et al. Case-based reasoning in IVF: Prediction and knowledge mining. Artif Intell Med 1998;12(1):1–24. DOI: 10.1016/s0933-3657(97)00037-7.

20. Hwang SW, Lee T, Kim H, et al. Classification of wood knots using artificial neural networks with texture and local feature-based image descriptors. Holzforschung 2022;76(1):1–13. DOI: 10.1515/hf-2021-0051.

21. Kharazmi E, Narliyeva L, Kalantar B, et al. Detection of cervical precancerous lesions using artificial intelligence: A systematic review and meta-analysis. J Lower Genital Tract Dis 2021;25(4):292–299. DOI: 10.1038/s41598-024-51880-4.

22. Sarno L, Neola D, Carbone L, et al. Use of artificial intelligence in obstetrics: not quite ready for prime time. Am J Obstet Gynecol MFM 2023;5(2):100792. DOI: 10.1016/j.ajogmf.2022.100792.

23. Harrer S, Shah P, Antony B, et al. Artificial intelligence for clinical trial design. Trends Pharmacol Sci 2019;40(8):577–591. DOI: 10.1016/j.tips.2019.05.005.

24. Smith EA, Horan WP, Demolle D, et al. Using artificial intelligence-based methods to address the placebo response in clinical trials. Innov Clin Neurosci 2022;19(1–3):60–70.

25. Zhou N, Zhang CT, Lv HY, et al. Concordance study between IBM Watson for oncology and clinical practice for patients with cancer in China. Oncologist 2019;24(6):812–819. DOI: 10.1634/theoncologist.2018-0255.

26. Vávra P, Roman J, Zonča P, et al. Recent development of augmented reality in surgery: A review. J Healthcare Eng 2017;2017:4574172. DOI: 10.1155/2017/4574172.

27. Angioni S, Pontis A, Nappi L, et al. Two-dimensional versus three-dimensional laparoscopy: A systematic review and meta-analysis in gynecology. J Minim Invasive Gynecol 2021;28(2):315–328. DOI: 10.1007/s13304-023-01465-z.

28. Song E, Yu F, Liu H, et al. A novel endoscope system for position detection and depth estimation of the ureter. J Med Syst 2016;40(12):266. DOI: 10.1007/s10916-016-0607-1.

29. Char DS, Shah NH, Magnus D. Implementing machine learning in health care: Addressing ethical challenges. New Eng J Med 2018;378(11):981–983. DOI: 10.1056/NEJMp1714229.

30. Ho CWL, Soon D, Caals K, et al. Governance of automated image analysis and artificial intelligence analytics in healthcare. Clin Radiol 2019;74(5):329–337. DOI: 10.1016/j.crad.2019.02.005.

31. Iftikhar P, Kuijpers MV, Khayyat A, et al. Artificial intelligence: A new paradigm in obstetrics and gynecology research and clinical practice. Cureus 2020;12(2):e7124. DOI: 10.7759/cureus.7124.

32. Hu C, Zhang W, Li P. 3D printing and its current status of application in obstetrics and gynecological diseases. Bioengineering (Basel) 2023;10(3):299. DOI: 10.3390/bioengineering10030299.

33. Khan B, Fatima H, Qureshi A, et al. Drawbacks of artificial intelligence and their potential solutions in the healthcare sector. Biomed Mater Devices 2023;1(8):3. DOI: 10.1007/s44174-023-00063-2.

34. Johnson A, Smith B, Williams C, et al. Challenges in ensuring AI algorithm quality and safety in clinical practice. J Med Artif Intell 12(3):215–230. DOI: 10.1186/s12909-023-04698-z.

35. Grzybowski A, Rao DP, Brona P, et al. Diagnostic accuracy of automated diabetic retinopathy image assessment softwares: IDx-DR and Medios Artificial Intelligence. Ophthalmic Res 2023;66(1):1286–1292. DOI: 10.1159/000534098.

36. Yoon JH, Kim EK. Deep learning-based artificial intelligence for mammography. Korean J Radiol 2021;22(8):1225–1239. DOI: 10.3348/kjr.2020.1210.

37. Yin J, Ngiam KY, Teo HH. Role of artificial intelligence applications in real-life clinical practice: Systematic review. J Med Internet Res 2021;23(4):e25759. DOI: 10.2196/25759.