Category: Rehab News

Una de las tareas del oncólogo es preparar a los pacientes con cáncer para tomar decisiones difíciles cuando se acerca el final. Sin embargo, no siempre se acuerdan de hacerlo.

En el sistema de salud de la Universidad de Pennsylvania (Penn Medicine), un algoritmo de inteligencia artificial (IA) que predice las probabilidades de muerte de los pacientes impulsa a los médicos a hablar con ellos sobre el tratamiento y sus preferencias al final de la vida.

Pero esta IA dista mucho de ser una herramienta fácil de usar, que se configura y listo.

Según un estudio de 2022, una revisión tecnológica de rutina descubrió que el algoritmo se había deteriorado durante la pandemia de covid, y que su precisión había bajado un 7% a la hora de predecir cuáles pacientes morirían.

Es probable que este deterioro haya tenido consecuencias concretas en la vida real. Ravi Parikh, oncólogo de la Universidad Emory y autor principal del estudio, explicó a KFF Health News que, en cientos de casos, la herramienta no logró alertar a los médicos para que comenzaran conversaciones cruciales con los pacientes, que podrían haberles evitado quimioterapias innecesarias.

Parikh cree que varios algoritmos diseñados para mejorar la atención médica se vieron afectados durante la pandemia, no sólo el de Penn Medicine. “Muchas instituciones no están monitoreando sistemáticamente el rendimiento de sus sistemas”, explicó.

Las fallas de los algoritmos son solo una parte de un dilema que los especialistas en informática y los médicos tienen desde hace tiempo, pero que ahora está empezando a desconcertar a los directivos de los hospitales y a los investigadores.

Los sistemas de inteligencia artificial requieren una supervisión continua y una dotación de personal altamente capacitado tanto para su implementación como para garantizar que funcionen bien.

En resumen: se necesitan más máquinas y más personas para asegurarse de que las nuevas herramientas no cometan errores.

“Todo el mundo piensa que la IA mejorará el acceso, aumentará la capacidad de los sistemas de salud y optimizará la atención, y eso suena muy bien”, dijo Nigam Shah, jefe de Datos Científicos en Stanford Health Care. “Pero, si el costo de la atención aumenta en un 20%, ¿es realmente viable?”.

A los funcionarios de gobierno les preocupa que los hospitales no tengan recursos para monitorear rigurosamente estas tecnologías. “He buscado por todas partes”, afirmó Robert Califf, comisionado de la Administración de Drogas y Alimentos (FDA), en una reciente mesa redonda sobre IA. “No creo que en Estados Unidos haya un solo sistema de salud que sea capaz de validar un algoritmo de IA implementado en un sistema de atención clínica”, agregó.

Sin embargo, la IA ya está ampliamente presente en el sector de la salud. Los algoritmos se usan para anticipar el riesgo de muerte o el deterioro de los pacientes, sugerir diagnósticos o clasificar la atención según la urgencia, registrar y resumir consultas para facilitar el trabajo de los médicos, e incluso para evaluar los reclamos de las aseguradoras.

Si los entusiastas de la tecnología están en lo cierto, la tecnología se volverá omnipresente… y rentable.

La empresa de inversión Bessemer Venture Partners ha identificado unas 20 startups de IA centradas en salud que están en vías de facturar $10 millones en un año cada una. La FDA ha aprobado cerca de mil productos de inteligencia artificial.

Evaluar si estas herramientas funcionan es todo un reto. Determinar si siguen funcionando bien —o si tienen fallas en sus sistemas operativos— es aún más complicado.

Por ejemplo, un estudio reciente de Yale Medicine analizó seis “sistemas de alerta precoz”, que avisan a los médicos cuándo es probable que un paciente se deteriore rápidamente.

Dana Edelson, médica de la Universidad de Chicago y cofundadora de una empresa que proporcionó un algoritmo para esta investigación, dijo que una supercomputadora revisó los datos durante varios días. El proceso fue fructífero, ya que mostró enormes diferencias de rendimiento entre los seis productos.

Para los hospitales y proveedores no es fácil seleccionar los mejores algoritmos en base a sus necesidades. No es habitual que los médicos tengan una supercomputadora a su disposición y no existe nada equiparable a un Consumer Reports para la IA.

“No tenemos normas”, aseguró Jesse Ehrenfeld, ex presidente de la Asociación Médica Estadounidense. “No existe nada que hoy se pueda señalar como una norma en relación con la forma de evaluar, supervisar o analizar el rendimiento de un modelo de algoritmo, con o sin inteligencia artificial, cuando se implementa”.

Quizás el producto de IA más común en las consultas médicas sea la “documentación ambiental”, un asistente tecnológico que escucha y transcribe las interacciones entre el médico y el paciente.

El año pasado, los inversores de Rock Health registraron un flujo de $353 millones en inversiones dirigidas hacia estas empresas de registros médicos. Pero, según Ehrenfeld, “actualmente no hay una pauta que permita comparar los resultados de estas herramientas”.

Esto es un problema, ya que incluso pequeños errores pueden ser devastadores. Un equipo de la Universidad de Stanford intentó usar grandes modelos lingüísticos —la tecnología que sustenta herramientas de IA populares como ChatGPT— para resumir el historial médico de los pacientes.

Más tarde, compararon los resultados con lo que hubiera escrito un médico.

“Incluso en el mejor de los casos, los modelos tenían una tasa de error del 35%”, explicó Shah, de Stanford. “Y en medicina, cuando estás escribiendo una historia clínica y te olvidas de una palabra, como por ejemplo ‘fiebre‘, se plantea un verdadero problema”, reflexionó.

A veces, las razones por las que los algoritmos fallan son bastante lógicas. Por ejemplo, las alteraciones en los datos estructurales pueden disminuir su efectividad, por ejemplo, cuando un hospital cambia de proveedor de laboratorio.

Sin embargo, en muchas otras ocasiones los problemas surgen sin un motivo aparente.

Sandy Aronson, ejecutivo tecnológico del programa de medicina personalizada del Mass General Brigham de Boston, contó que cuando su equipo probó una aplicación destinada a ayudar a los consejeros en genética a localizar bibliografía relevante sobre variantes del ADN, el producto sufrió “no determinismo”. Esto significa que, cuando se le hacía varias veces la misma pregunta en un breve período de tiempo, daba resultados diferentes.

Aronson está entusiasmado con el potencial de los grandes modelos lingüísticos para resumir conocimientos que simplifiquen el trabajo de los sobrecargados consejeros, pero considera que “la tecnología tiene que mejorar”.

Si hay pocas métricas y estándares, y los errores pueden surgir por razones raras, ¿qué deben hacer las instituciones? Invertir en una gran cantidad de recursos. En Stanford, Shah comentó que les llevó entre ocho y diez meses revisar solo dos modelos en términos de equidad y confiabilidad.

Expertos entrevistados por KFF Health News plantearon la idea de que la inteligencia artificial supervise a la inteligencia artificial, y que algún genio (humano) en datos supervise a ambas.

Todos reconocieron que esto requeriría que las organizaciones gastaran aún más dinero, una pretensión difícil de satisfacer dada la realidad de los presupuestos hospitalarios y la limitada oferta de especialistas en tecnología de IA.

“Es estupendo tener una perspectiva en la que estamos haciendo un esfuerzo colosal para poder monitorear un modelo con otro modelo”, dijo Shah. “Pero ¿es eso realmente lo que se quería? ¿Cuánta gente más vamos a necesitar?”.

Esta historia fue producida por KFF Health News, conocido antes como Kaiser Health News (KHN), una redacción nacional que produce periodismo en profundidad sobre temas de salud y es uno de los principales programas operativos de KFF, la fuente independiente de investigación de políticas de salud, encuestas y periodismo. 

The Host

The new, GOP-led, 119^th Congress and President-elect Donald Trump have big legislative plans for the year — which mostly don’t include health policy. But health is likely to play an important supporting role in efforts to renew tax cuts, revise immigration policies, and alter trade — if only to help pay for some Republican initiatives.

Meanwhile, the outgoing Biden administration is racing to finish its health policy to-do list, including finalizing a policy that bars credit bureaus from including medical debt on individuals’ credit reports.

This week’s panelists are Julie Rovner of KFF Health News, Shefali Luthra of The 19^th, Alice Miranda Ollstein of Politico, and Lauren Weber of The Washington Post.

Among the takeaways from this week’s episode:

• The 119th Congress is now in session. Health care doesn’t make the list of priorities as lawmakers lay the table for the incoming Trump administration — though Republicans have floated Medicaid work requirements to cut federal spending.
• A lot of health legislation hit the cutting-room floor in December, including a bipartisan proposal targeting pharmacy benefit managers — which would have saved the federal government and patients billions of dollars. And speaking of bipartisan efforts, a congressional report from the Senate Budget Committee adds to evidence that private equity involvement in care is associated with worse outcomes for patients — notably, lawmakers’ constituents.
• As the nation bids a final farewell to former President Jimmy Carter, his global health work, in particular, is being celebrated — especially his efforts to eradicate such devastating diseases as Guinea worm disease and river blindness.
• Meanwhile, the Biden administration finalized the rule barring medical debt from appearing on credit reports. The surgeon general cautions that alcohol should come with warning labels noting cancer risk. And the new Senate Republican leader is raising abortion-related legislation to require lifesaving care for all babies born alive — yet those protections already exist.

Plus, for “extra credit,” the panelists suggest health policy stories they read this week that they think you should read, too:

Julie Rovner: The Wall Street Journal’s “UnitedHealth’s Army of Doctors Helped It Collect Billions More From Medicare,” by Christopher Weaver, Anna Wilde Mathews, and Tom McGinty.

Alice Miranda Ollstein: The New York Times’ “Ozempic, Lego Bricks and Hearing Aids: What Trump’s Greenland Plan Could Hit,” by Ana Swanson and Jenny Gross.

Shefali Luthra: Vox.com’s “Gigantic SUVs Are a Public Health Threat. Why Don’t We Treat Them Like One?” by David Zipper.

Lauren Weber: The Washington Post’s “Laws Restrict U.S. Shipping of Vape Products. Many Companies Do It Anyway,” by David Ovalle and Rachel Roubein.

Also mentioned in this week’s podcast:

The Senate Budget Committee’s “Profits Over Patients: The Harmful Effects of Private Equity on the U.S. Health Care System.”

To hear all our podcasts, click here.

And subscribe to KFF Health News’ “What the Health?” on Spotify, Apple Podcasts, Pocket Casts, or wherever you listen to podcasts.

This article was produced by KFF Health News, a national newsroom that produces in-depth journalism about health issues and is one of the core operating programs at KFF — the independent source for health policy research, polling, and journalism. 

We’ve all read the stories and seen the images: The life-threatening heat waves. The wildfires of unprecedented ferocity. The record-breaking storms washing away entire neighborhoods. The melting glaciers, the rising sea levels, the coastal flooding.

As California wildfires stretch into the colder months and hurricane survivors sort through the ruins left by floodwaters, let’s talk about an underreported victim of climate change: the emotional well-being of young people.

A nascent but growing body of research shows that a large proportion of adolescents and young adults, in the United States and abroad, feel anxious and worried about the impact of an unstable climate in their lives today and in the future.

Abby Rafeek, 14, is disquieted by the ravages of climate change, both near her home and far away. “It’s definitely affecting my life, because it’s causing stress thinking about the future and how, if we’re not addressing the problem now as a society, our planet is going to get worse,” says Rafeek, a high school student who lives in Gardena, California, a city of 58,000 about 15 miles south of downtown Los Angeles.

She says wildfires are a particular worry for her. “That’s closer to where I live, so it’s a bigger problem for me personally, and it also causes a lot of damage to the surrounding areas,” she says. “And also, the air gets messed up.”

In April, Rafeek took a survey on climate change for kids ages 12-17 during a visit to the emergency room at Children’s Hospital of Orange County.

Rammy Assaf, a pediatric emergency physician at the hospital, adapted the survey from one developed five years ago for adults. He administered his version last year to over 800 kids ages 12-17 and their caregivers. He says initial results show climate change is a serious cause of concern for the emotional security and well-being of young people.

Assaf has followed up with the kids to ask more open-ended questions, including whether they believe climate change will be solved in their lifetimes; how they feel when they read about extreme climate events; what they think about the future of the planet; and with whom they are able to discuss their concerns.

“When asked about their outlook for the future, the first words they will use are helpless, powerless, hopeless,” Assaf says. “These are very strong emotions.”

Assaf says he would like to see questions about climate change included in mental health screenings at pediatricians’ offices and in other settings where children get medical care. The American Academy of Pediatrics recommends that counseling on climate change be incorporated into the clinical practice of pediatricians and into medical school curriculums, but not with specific regard to mental health screening.

Assaf says anxiety about climate change intersects with the broader mental health crisis among youth, which has been marked by a rise in depression, loneliness, and suicide over the past decade, though there are recent signs it may be improving slightly.

A 2022 Harris Poll of 1,500 U.S. teenagers found that 89% of them regularly think about the environment, “with the majority feeling more worried than hopeful.” In addition, 69% said they feared they and their families would be affected by climate change in the near future. And 82% said they expected to have to make key life decisions — including where to live and whether to have children — based on the state of the environment.

And the impact is clearly not limited to the U.S. A 2021 survey of 10,000 16- to 25-year-olds across 10 countries found “59% were very or extremely worried and 84% were at least moderately worried” about climate change.

Susan Clayton, chair of the psychology department at the College of Wooster in Ohio, says climate change anxiety may be more pronounced among younger people than adults. “Older adults didn’t grow up being as aware of climate change or thinking about it very much, so there’s still a barrier to get over to accept it’s a real thing,” says Clayton, who co-created the adult climate change survey that Assaf adapted for younger people.

By contrast, “adolescents grew up with it as a real thing,” Clayton says. “Knowing you have the bulk of your life ahead of you gives you a very different view of what your life will be like.” She adds that younger people in particular feel betrayed by their government, which they don’t think is taking the problem seriously enough, and “this feeling of betrayal is associated with greater anxiety about the climate.”

Rafeek believes climate change is not being addressed with sufficient resolve. “I think if we figure out how to live on Mars and explore the deep sea, we could definitely figure out how to live here in a healthy environment,” she says.

If you are a parent whose children show signs of climate anxiety, you can help.

Louise Chawla, professor emerita in the environmental design program at the University of Colorado-Boulder, says the most important thing is to listen in an open-ended way. “Let there be space for kids to express their emotions. Just listen to them and let them know it’s safe to express these emotions,” says Chawla, who co-founded the nonprofit Growing Up Boulder, which works with the city’s schools to encourage kids to engage civically, including to help shape their local environment.

Chawla and others recommend family activities that reinforce a commitment to the environment. They can be as simple as walking or biking and participating in cleanup or recycling efforts. Also, encourage your children to join activities and advocacy efforts sponsored by environmental, civic, or religious organizations.

Working with others can help alleviate stress and feelings of powerlessness by reassuring kids they are not alone and that they can be proactive.

Worries about climate change should be seen as a learning opportunity that might even lead some kids to their life’s path, says Vickie Mays, professor of psychology and health policy at UCLA, who teaches a class on climate change and mental health — one of eight similar courses offered recently at UC campuses.

“We should get out of this habit of ‘everything’s a mental health problem,’” Mays says, “and understand that often a challenge, a stress, a worry can be turned into advocacy, activism, or a reach for new knowledge to change the situation.”

The federal Consumer Financial Protection Bureau on Tuesday issued new regulations barring medical debts from American credit reports, enacting a major new consumer protection just days before President Joe Biden is set to leave office.

The rules ban credit agencies from including medical debts on consumers’ credit reports and prohibit lenders from considering medical information in assessing borrowers.

These rules, which the federal watchdog agency proposed in June, could be reversed after President-elect Donald Trump takes office Jan. 20. But by finalizing the regulations now, the CFPB effectively dared the incoming Trump administration and its Republican allies in Congress to undue rules that are broadly popular and could help millions of people who are burdened by medical debt.

“People who get sick shouldn’t have their financial future upended,” CFPB Director Rohit Chopra said in announcing the new rules. “The CFPB’s final rule will close a special carveout that has allowed debt collectors to abuse the credit reporting system to coerce people into paying medical bills they may not even owe.”

The regulations fulfill a pledge by the Biden administration to address the scourge of health care debt, a problem that touches an estimated 100 million Americans, forcing many to make sacrifices such as limiting food, clothing, and other essentials.

Credit reporting, a threat that has been wielded by medical providers and debt collectors to get patients to pay their bills, is the most common collection tactic used by hospitals, a KFF Health News analysis found.

The impact can be devastating, especially for those with large health care debts.

There is growing evidence, for example, that credit scores depressed by medical debt can threaten people’s access to housing and drive homelessness. People with low credit scores can also have trouble getting a loan or can be forced to borrow at higher interest rates.

That has prompted states including Colorado, New York, and California to enact legislation prohibiting medical debt from being included on residents’ credit reports or factored into their credit scores. Still, many patients and consumer advocates have pushed for a national ban.

The CFPB has estimated that the new credit reporting rule will boost the credit scores of people with medical debt on their credit reports by an average of 20 points.

But the agency’s efforts to restrict medical debt collections have drawn fierce pushback from the collections industry. And the new rules will almost certainly be challenged in court.

Congressional Republicans have frequently criticized the watchdog agency. Last year, then-chair of the House Financial Services Committee Patrick McHenry (R-N.C.) labeled the CFPB’s medical debt proposal “regulatory overreach.”

More recently, billionaire Elon Musk, whom Trump has tapped to co-lead his initiative to shrink government, called for the elimination of the watchdog agency. “Delete CFPB,” Musk posted on the social platform X.

This article was produced by KFF Health News, a national newsroom that produces in-depth journalism about health issues and is one of the core operating programs at KFF — the independent source for health policy research, polling, and journalism.