S. Sawan, N. Eftekhari, K. Linton-Reid, N. Wood, T. A. Numan, E. O. Aboagye, C. Angione, the Royal College of Obstetricians and Gynaecologists
<p>The term artificial intelligence (AI) is believed to have been coined by John McCarthy et al. at the Dartmouth Summer Research Project in 1956, when it was proposed that a machine can be made to simulate ‘every aspect of learning or any other features of intelligence’ [<span>1</span>]. AI is a rapidly evolving field with expanding potentials that is increasingly becoming an integral part of our daily lives. Everyday examples include internet search engines, recommended posts on social media, financial sector forecast, disease outbreak modelling, defence and weaponry, and even the editing of medical articles [<span>2-4</span>].</p><p>While there is no universally agreed definition, AI can refer to a branch of informatics that engineer computer systems capable of performing tasks that typically require human intelligence such as reasoning, adaptation and learning via feedback processes [<span>5, 6</span>]. The National Institute for Health and Care Excellence (NICE) has noted that the exact definition of AI in healthcare could be context-dependent and that the extent of AI incorporation into digital health technologies could vary widely [<span>7, 8</span>].</p><p>In AI, computer systems are built using algorithms—which are sets of mathematical instructions constructed by coding engineers—to uncover patterns and relationships among variables by mining and mapping data and then selecting the best model for a specified purpose [<span>6, 9, 10</span>]. Algorithms in AI are designed so they can learn and, hence, refine their own performance. This is unlike conventional algorithms used in traditional computing, which are engineered to follow predefined strict instructions and rules with no inherent capability for learning or performance improvement [<span>11</span>]. Generally, AI algorithms are trained on a dataset (called training data) and then are tested to assess performance on another unseen dataset (testing data) prior to implementation on external or validation data. Typically, both training data and testing data are obtained from the same dataset, which is usually divided according to a specified ratio and allocation method [<span>6</span>].</p><p>This guidance is for healthcare professionals who care for women, non-binary and trans people. Within this document we use the terms woman and women's health. However, it is important to acknowledge that it is not only women for whom it is necessary to access women's health and reproductive services in order to maintain their gynaecological health and reproductive wellbeing. Gynaecological and obstetric services and delivery of care must therefore be appropriate, inclusive and sensitive to the needs of those individuals whose gender identity does not align with the sex recorded at birth.</p><p>The contribution of AI in healthcare is widely celebrated on social and traditional media platforms. It is regarded as an example of good use and a positive role for AI in the face of growing concerns among AI
{"title":"Artificial Intelligence in Gynaecology Oncology","authors":"S. Sawan, N. Eftekhari, K. Linton-Reid, N. Wood, T. A. Numan, E. O. Aboagye, C. Angione, the Royal College of Obstetricians and Gynaecologists","doi":"10.1111/1471-0528.70005","DOIUrl":"10.1111/1471-0528.70005","url":null,"abstract":"<p>The term artificial intelligence (AI) is believed to have been coined by John McCarthy et al. at the Dartmouth Summer Research Project in 1956, when it was proposed that a machine can be made to simulate ‘every aspect of learning or any other features of intelligence’ [<span>1</span>]. AI is a rapidly evolving field with expanding potentials that is increasingly becoming an integral part of our daily lives. Everyday examples include internet search engines, recommended posts on social media, financial sector forecast, disease outbreak modelling, defence and weaponry, and even the editing of medical articles [<span>2-4</span>].</p><p>While there is no universally agreed definition, AI can refer to a branch of informatics that engineer computer systems capable of performing tasks that typically require human intelligence such as reasoning, adaptation and learning via feedback processes [<span>5, 6</span>]. The National Institute for Health and Care Excellence (NICE) has noted that the exact definition of AI in healthcare could be context-dependent and that the extent of AI incorporation into digital health technologies could vary widely [<span>7, 8</span>].</p><p>In AI, computer systems are built using algorithms—which are sets of mathematical instructions constructed by coding engineers—to uncover patterns and relationships among variables by mining and mapping data and then selecting the best model for a specified purpose [<span>6, 9, 10</span>]. Algorithms in AI are designed so they can learn and, hence, refine their own performance. This is unlike conventional algorithms used in traditional computing, which are engineered to follow predefined strict instructions and rules with no inherent capability for learning or performance improvement [<span>11</span>]. Generally, AI algorithms are trained on a dataset (called training data) and then are tested to assess performance on another unseen dataset (testing data) prior to implementation on external or validation data. Typically, both training data and testing data are obtained from the same dataset, which is usually divided according to a specified ratio and allocation method [<span>6</span>].</p><p>This guidance is for healthcare professionals who care for women, non-binary and trans people. Within this document we use the terms woman and women's health. However, it is important to acknowledge that it is not only women for whom it is necessary to access women's health and reproductive services in order to maintain their gynaecological health and reproductive wellbeing. Gynaecological and obstetric services and delivery of care must therefore be appropriate, inclusive and sensitive to the needs of those individuals whose gender identity does not align with the sex recorded at birth.</p><p>The contribution of AI in healthcare is widely celebrated on social and traditional media platforms. It is regarded as an example of good use and a positive role for AI in the face of growing concerns among AI","PeriodicalId":50729,"journal":{"name":"Bjog-An International Journal of Obstetrics and Gynaecology","volume":"133 4","pages":"e1-e17"},"PeriodicalIF":4.3,"publicationDate":"2025-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://obgyn.onlinelibrary.wiley.com/doi/epdf/10.1111/1471-0528.70005","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145288246","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
G. Justus Hofmeyr, Mandisa Singata-Madliki, Sara Della Ripa, Andrew D. Weeks
The global impact of life-saving medical devices is directly related to their availability. Access may be limited by cost, availability, or lack of information regarding effectiveness and safety. Addressing the inequity in access requires concerted effort from device developers, the research community, global agencies and professional organisations. We discuss, with examples, three strategies to promote equity: low-cost, purpose-built innovation, improvisation and off-label use. First, developing simple, safe and low-cost innovative devices can be an effective way of increasing global access. For example, the BabySaver Kit facilitates intact-cord neonatal bedside resuscitation. Re-usability is an important design feature for both cost and environment, exemplified by the MaternaWell tray for blood loss monitoring after birth. A second strategy is improvisation using commonly available hospital items. This can extend device availability into settings where purpose-designed devices are unavailable or unaffordable. Examples include the use of condoms or glove balloons for uterine balloon tamponade (UBT) to treat postpartum haemorrhage (PPH), elastic catheters for uterine tourniquet, and plastic tubing for posterior axilla sling traction in shoulder dystocia. However, the lack of systematically developed evidence and governance approvals can lead to wide variation in training, technique, and device specifications. Finally, some of these quality issues are addressed by using approved medical devices ‘off-label.’ However, they can have similar problems of variation in technique and depend on the uncoordinated efforts of researchers and clinicians to generate an evidence base. Examples include the Foley catheter for labour induction and the Levin stomach tube for suction tube uterine tamponade for PPH. WHO has pathways to facilitate global access to important public health device innovations. Global agencies and professional organisations also have a major role to play in providing co-ordination, platforms for data sharing, practice guidelines, instructions for use on off-label devices and robust data on their safety and effectiveness.
{"title":"Achieving Equitable Access to Obstetric Devices Through Innovation, Improvisation and Off-Label Use","authors":"G. Justus Hofmeyr, Mandisa Singata-Madliki, Sara Della Ripa, Andrew D. Weeks","doi":"10.1111/1471-0528.70058","DOIUrl":"10.1111/1471-0528.70058","url":null,"abstract":"<p>The global impact of life-saving medical devices is directly related to their availability. Access may be limited by cost, availability, or lack of information regarding effectiveness and safety. Addressing the inequity in access requires concerted effort from device developers, the research community, global agencies and professional organisations. We discuss, with examples, three strategies to promote equity: low-cost, purpose-built innovation, improvisation and off-label use. First, developing simple, safe and low-cost innovative devices can be an effective way of increasing global access. For example, the BabySaver Kit facilitates intact-cord neonatal bedside resuscitation. Re-usability is an important design feature for both cost and environment, exemplified by the MaternaWell tray for blood loss monitoring after birth. A second strategy is improvisation using commonly available hospital items. This can extend device availability into settings where purpose-designed devices are unavailable or unaffordable. Examples include the use of condoms or glove balloons for uterine balloon tamponade (UBT) to treat postpartum haemorrhage (PPH), elastic catheters for uterine tourniquet, and plastic tubing for posterior axilla sling traction in shoulder dystocia. However, the lack of systematically developed evidence and governance approvals can lead to wide variation in training, technique, and device specifications. Finally, some of these quality issues are addressed by using approved medical devices ‘off-label.’ However, they can have similar problems of variation in technique and depend on the uncoordinated efforts of researchers and clinicians to generate an evidence base. Examples include the Foley catheter for labour induction and the Levin stomach tube for suction tube uterine tamponade for PPH. WHO has pathways to facilitate global access to important public health device innovations. Global agencies and professional organisations also have a major role to play in providing co-ordination, platforms for data sharing, practice guidelines, instructions for use on off-label devices and robust data on their safety and effectiveness.</p>","PeriodicalId":50729,"journal":{"name":"Bjog-An International Journal of Obstetrics and Gynaecology","volume":"132 13","pages":"1903-1909"},"PeriodicalIF":4.3,"publicationDate":"2025-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://obgyn.onlinelibrary.wiley.com/doi/epdf/10.1111/1471-0528.70058","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145282723","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}