Science Quick Takes: Flu Surge, Sleep-FM Predicts Disease, CES 2026 AI, and Woodpecker Biomechanics

Short summary

In this week’s Science Quickly, the team reports a record flu season in the US, noting high visits for respiratory illness and hospitalizations, with the H3N2 subclade K driving much of the surge. Vaccination guidance for children has shifted to doctor-patient discussions, and public health messaging is revisited in light of policy changes. The episode also introduces Sleep FM, a foundation model trained on polysomnography data to predict health outcomes such as Parkinson's, dementia, heart disease, and cancer, with best predictions arising from combined data streams. The CES 2026 segment highlights AI in physical devices and even Stevie Wonder’s stance on AI in music. The show closes with woodpecker biomechanics that reveal the body-wide nature of their pecking.

Influenza surge and vaccine guidance

The episode opens by describing a historically high rate of respiratory illness visits to health care providers in the US for the week ending December 27, as reported by the CDC, marking the highest level since tracking began in 1997. So far this season, flu activity has contributed to roughly 120,000 hospitalizations and about 5,000 deaths, including nine children. Public health guidance on flu vaccination has shifted, with the CDC recommending that parents discuss vaccination with their child’s doctor rather than a universal mandate for all children over six months old. The segment notes that a new H3N2 subclade, K, appears to drive much of the uptick, though vaccines still provide protection against severe illness and death. In addition, wearing well-fitted masks in indoor public spaces remains a practical precaution. The discussion also touches on past CDC vaccination campaigns and the role of public messaging in vaccination uptake. Quotes will appear in context below

“The good news is that vaccinated people are less likely to be hospitalized or die from the flu” — CDC researchers

Sleep FM foundation model and disease risk prediction

Following the flu news, the podcast shifts to a Stanford-led study on a foundation model named Sleep FM trained on polysomnography data, the gold standard in sleep assessment. Sleep FM was trained on nearly 600,000 hours of sleep data from 65,000 patients, enabling it to analyze sleep stages and other physiological signals. The researchers tested Sleep FM's ability to predict long-term health outcomes and found it successfully linked sleep behavior to risks including Parkinson’s disease, dementia, hypertensive heart disease, heart attacks, prostate cancer, breast cancer, and death. A key finding was that the strongest predictors emerged when integrating multi-modal data, such as a brain that appears asleep with an awake heart, illustrating the potential of holistic, cross-system modeling. Emmanuel Mignon, a co-senior author, emphasized that the integration of data streams provides better predictive power than single-signal analyses. The segment closes by noting Sleep FM’s implications for health monitoring and disease risk stratification. Quotes will appear in context below

“a brain that looks asleep, but a heart that looks awake, for example, seemed to spell trouble” — Emmanuel Mignon, co-senior author

CES 2026: AI moving from chat to the real world

Eric Sullivan reports from Las Vegas on CES 2026, describing AI as infrastructure that now permeates physical products, devices, and humanoid robotics. The coverage highlights AI-enabled hardware, enhanced accessibility tech, and tangible demonstrations where AI-enabled systems operate alongside humans. A notable moment in the interview was Stevie Wonder’s stance on AI in music, where he stated he would not let AI program or replace his artistic process, signaling a nuanced view of human-centered creativity amid rapid AI development. Sullivan concludes with a takeaway that AI is becoming a core component of computing platforms and devices, bridging software and hardware to shape consumer technology in the coming year. Quotes will appear in context below

“I will not let my music be programmed. I'm not going to use it to do me and do the music I've done” — Stevie Wonder

Woodpecker biomechanics reveal nature’s hammer

The final segment covers Brown University research on downy woodpeckers, which can strike wood with forces up to 30 times their body weight and up to 13 pecks per second. The study involved humane capture and in-lab measurement of muscle activity, showing pecking is a full-body action with coordinated use of tails, abdominal muscles, hip flexors, and neck muscles to absorb impact. Lead author Nicholas Anderson described woodpeckers as nature’s hammers, highlighting the biomechanics that protect the birds from brain injury during rapid pecking. The research underscores the remarkable integration of anatomy and behavior that enables high-impact, repetitive woodpecking. Quotes will appear in context below

“Woodpeckers really are nature's hammer in a sense” — Nicholas Anderson, Brown University biologist