AI-automated referral for patients with visual impairment

Alam, Minhaj Nur; Hallak, Joelle

doi:10.1016/s2589-7500(20)30286-7

Cited by 3 publications

(2 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…(el United States Medical Licensing Exam -USMLE-), ya que ChatGPT-4 contestó correctamente el 90% de las preguntas . Esta competencia de las IA indica un potencial interés para satisfacer demandas profesionales en la práctica de la medicina, especialmente cuando se manejan bases de datos con miles de pacientes 10 o cuando se está ante un caso clínico complejo 11 , habiendo mostrado sus potenciales capacidades en el terreno de la medicina forense 12 , salud pública 13 , farmacología 14 , oftalmología 15 , cardiología 16 , oncología 17 , neumología 18 , neurología 19 , neurocirugía 20 , hepatología 21 , odontología 22 , y angiología 23 , entre otras especialidades 24 . Pero no hay que olvidar que los sistemas de IA estudiados, dotados de un lenguaje natural en base al enorme volumen de información disponible, no son infalibles y cometen errores derivados de la ausencia de comprensión del texto que proponen, lo que podría llevar a una mala praxis 25 .…”

Section: Discussionunclassified

Assessing 4 artificial intelligence systems’ knowledge of a subspecialty of emergency medicine: clinical toxicology

Nogué-Xarau,

Amigó-Tadín,

Ríos-Guillermo

2024

Rev Esp Urg Emerg

View full text Add to dashboard Cite

OBJETIVO. La inteligencia artificial (IA) es una disciplina de la informática que se encarga de crear sistemas capaces de realizar tareas que se atribuyen a la inteligencia humana. El objetivo principal de este estudio ha sido evaluar las respuestas de algunas IA a preguntas del campo de la toxicología clínica (TC). MATERIAL Y MÉTODOS. Se han valorado cuatro aplicaciones de IA: ChatGPT, Bing, LuzIA y Bard. Para evaluar sus conocimientos en TC se les formularon 30 preguntas sobre diversos aspectos de la TC. Cada pregunta ofrecía cinco opciones de respuesta, de las cuales sólo una era correcta. Se evaluó el acierto/error en la respuesta, así como si había apoyo bibliográfico. Si se detectaban respuestas erróneas, se reformuló la misma pregunta, pero utilizando otra forma de lenguaje para evaluar de nuevo la respuesta y ver si la misma era sensible a la calidad de la pregunta. Los datos se introdujeron en una base SPSS para su análisis estadístico. Se consideró significativo un valor de p < 0,05. RESULTADOS. Los porcentajes de respuestas acertadas fueron del 70% (Bing), 67% (ChatGPT y LuzIA) y 57% (Bard), sin diferencias estadísticamente significativas. Al reformular las preguntas en los casos en los que la respuesta de la IA había sido errónea, los porcentajes de aciertos subieron en los cuatro sistemas, pero sin diferencias significativas. En sus respuestas, Bing ofreció el acceso directo a tres citas bibliográficas y Bard a cuatro, pero su presencia en PubMed era muy baja (7,2% y 0,85% respectivamente). CONCLUSIONES. Los cuatro sistemas de IA han mostrado una capacidad de acierto en más del 50% de las preguntas formuladas de TC. No obstante, el soporte bibliográfico que proporcionan es escaso y de muy baja calidad.

show abstract

Section: Discussionunclassified

Assessing 4 artificial intelligence systems’ knowledge of a subspecialty of emergency medicine: clinical toxicology

Nogué-Xarau,

Amigó-Tadín,

Ríos-Guillermo

2024

Rev Esp Urg Emerg

View full text Add to dashboard Cite

show abstract

“…The challenge for telemedicine is to put emphasis on preventive medicine since the distance will no longer be a factor between patient and health professional. According to several studies, AI has been used in automatic screening of diabetic retinopathy [55], coordination and management of breast cancer [56], clinical deployment of prostate cancer diagnosis [57], detection and classification of chronic noncommunicable diseases [58], heart disease prediction [59], pneumonia detection [60], health screening [61], urinalysis examination [62], classification of skin diseases [63], and referral system [64] and voice controlled navigational aid [65] for visually impaired patients. Based on the studies, telemedicine can use AI for the purpose of diagnostics, communication of cure and medication, and monitoring of patients after surgery on the way to recovery or telerehabilitation after the pandemic [20], and before the pandemic [66].…”

Section: ) Artificial Intelligencementioning

confidence: 99%

Telemedicine Implementation Challenges in Underserved Areas of the Philippines

Cruz¹,

Tolentino²

2021

IJETAE

View full text Add to dashboard Cite

During these difficult times of global pandemic, mobility restrictions and lockdowns were put in place to prevent an outbreak. However, patients with medical needs must deal with regular face-to-face consultations with physicians for their prescriptions and medications. It is important to review the various telemedicine research in the Philippines and abroad and how to formulate a national framework for telemedicine so the program can be implemented in the entire country. As a developing nation and an archipelago, the Philippines needs to work on massive infrastructure project for national broadband and health care facilities. The increase of medical professionals per 1000 of population and prevent them from migrations will help implement such innovation in medical field. A national policy to support the program in terms of funding and its transparency, establishment of processes and standards, and safeguarding digital information should be put in place. The country will need to address these gaps in technology and processes and encourage the population to partake in national health plans and initiatives to increase awareness and engagement of all stakeholders. Keywords— Challenges, Developing Countries, eHealth, Implementation, IoMT, Philippines, Rural, Telehealth, Telemedicine, Underserved

show abstract

Algorithmic legitimacy in clinical decision-making

Holm

2023

Ethics Inf Technol

View full text Add to dashboard Cite

Machine learning algorithms are expected to improve referral decisions. In this article I discuss the legitimacy of deferring referral decisions in primary care to recommendations from such algorithms. The standard justification for introducing algorithmic decision procedures to make referral decisions is that they are more accurate than the available practitioners. The improvement in accuracy will ensure more efficient use of scarce health resources and improve patient care. In this article I introduce a proceduralist framework for discussing the legitimacy of algorithmic referral decisions and I argue that in the context of referral decisions the legitimacy of an algorithmic decision procedure can be fully accounted for in terms of the instrumental values of accuracy and fairness. I end by considering how my discussion of procedural algorithmic legitimacy relates to the debate on algorithmic fairness.

show abstract

AI-automated referral for patients with visual impairment

Cited by 3 publications

References 11 publications

Assessing 4 artificial intelligence systems’ knowledge of a subspecialty of emergency medicine: clinical toxicology

Assessing 4 artificial intelligence systems’ knowledge of a subspecialty of emergency medicine: clinical toxicology

Telemedicine Implementation Challenges in Underserved Areas of the Philippines

Algorithmic legitimacy in clinical decision-making

Contact Info

Product

Resources

About