Voice features such as tone, clarity could help detect early signs of laryngeal cancer: Researchers
Findings published in the journal Frontiers in Digital Health could help in developing artificial intelligence (AI) models for detecting abnormalities, or 'lesions', in vocal folds from the sound of a voice -- lesions may be benign, but could also represent early stages of laryngeal cancer, the researchers said.
Current diagnostic procedures, such as endoscopies and biopsies, are invasive, they added.
"Here we show that with this dataset (of voice recordings) we could use
vocal biomarkers
to distinguish voices from patients with vocal fold lesions from those without such lesions," said author Dr Phillip Jenkins, a postdoctoral fellow in clinical informatics at Oregon Health and Science University, US.
The team analysed tone, pitch, volume, and clarity of over 12,500 voice recordings of 306 participants, taken from the publicly available 'Bridge2AI-Voice' dataset.
"Among the overall sample, significant differences were identified in HNR (harmonic-to-noise ratio) and fundamental frequency between benign lesions and both healthy controls and laryngeal cancer," the authors wrote. A 'harmonic-to-noise' ratio, or 'clarity', measures the amount of noise in a voice signal.
While marked differences in voice features were found between men having benign lesions in vocal folds, and those with laryngeal cancer, the researchers could not find distinct acoustic features among women's voices.
It is possible that a larger dataset would reveal such differences, they said.
Changes in the harmonic-to-noise ratio can be helpful in monitoring how lesions in vocal folds evolve over time and detecting laryngeal cancer at an early stage, at least in men, the researchers said.
"Our results suggest that ethically sourced, large, multi-institutional datasets like Bridge2AI-Voice could soon help make our voice a practical biomarker for cancer risk in clinical care," said Dr Jenkins.
The team is looking to train the AI model on larger datasets of voice recordings and test it among men and women.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Time of India
a day ago
- Time of India
Voice features such as tone, clarity could help detect early signs of laryngeal cancer: Researchers
New Delhi: Researchers have found that voice features in recordings -- such as tone, pitch and clarity -- could help detect early warning signs of cancer of the larynx (voice box). Findings published in the journal Frontiers in Digital Health could help in developing artificial intelligence (AI) models for detecting abnormalities, or 'lesions', in vocal folds from the sound of a voice -- lesions may be benign, but could also represent early stages of laryngeal cancer, the researchers said. Current diagnostic procedures, such as endoscopies and biopsies, are invasive, they added. "Here we show that with this dataset (of voice recordings) we could use vocal biomarkers to distinguish voices from patients with vocal fold lesions from those without such lesions," said author Dr Phillip Jenkins, a postdoctoral fellow in clinical informatics at Oregon Health and Science University, US. The team analysed tone, pitch, volume, and clarity of over 12,500 voice recordings of 306 participants, taken from the publicly available 'Bridge2AI-Voice' dataset. "Among the overall sample, significant differences were identified in HNR (harmonic-to-noise ratio) and fundamental frequency between benign lesions and both healthy controls and laryngeal cancer," the authors wrote. A 'harmonic-to-noise' ratio, or 'clarity', measures the amount of noise in a voice signal. While marked differences in voice features were found between men having benign lesions in vocal folds, and those with laryngeal cancer, the researchers could not find distinct acoustic features among women's voices. It is possible that a larger dataset would reveal such differences, they said. Changes in the harmonic-to-noise ratio can be helpful in monitoring how lesions in vocal folds evolve over time and detecting laryngeal cancer at an early stage, at least in men, the researchers said. "Our results suggest that ethically sourced, large, multi-institutional datasets like Bridge2AI-Voice could soon help make our voice a practical biomarker for cancer risk in clinical care," said Dr Jenkins. The team is looking to train the AI model on larger datasets of voice recordings and test it among men and women.
Economic Times
25-07-2025
- Economic Times
AI is being used to design custom proteins that help immune cells find and kill cancer
Live Events (You can now subscribe to our (You can now subscribe to our Economic Times WhatsApp channel Scientists have used artificial intelligence to design custom proteins that act like a 'GPS' for the immune system, guiding T cells directly to cancer cells and helping them destroy tumors with remarkable new research, published Thursday(July 24) in the journal Science, demonstrates how generative AI can create tiny proteins that dramatically enhance the cancer-killing power of human immune cells. In lab tests, T cells outfitted with these proteins swiftly located and killed melanoma cells, a fast-spreading and often deadly form of skin cancer.'It's like giving immune cells Google Maps for cancer,' said lead researcher Dr. Timothy Jenkins, a medical biotechnologist at the Technical University of Denmark. 'We're helping T cells find the tumor much more efficiently than they could on their own.'The work is an early-stage form of immunotherapy, a treatment approach that boosts the body's natural defenses to fight disease. The idea builds on successful techniques like CAR T-cell therapy but adds a major twist: Instead of relying on naturally occurring cell receptors, which can take months to isolate and test, the scientists used AI to design new proteins from scratch in just use of AI toolsTo create the molecular guides, the researchers used three AI tools. First, a generative model called RFdiffusion was used to analyze the structure of a specific cancer target known as NY-ESO-1, a protein found on many tumor cells. Then, another model proposed amino acid sequences that would fold into the right shape to bind the target. A third model helped narrow tens of thousands of options down to just 44, which were tested in the lab. One design stood out, and it study builds on a wave of recent breakthroughs in computational biology, including technologies that led to the 2024 Nobel Prize in Chemistry for protein structure prediction. Earlier this year, Jenkins' team also developed AI-designed proteins for improved snakebite trials are still years away, and many questions remain about how the proteins will perform in the human body. But the researchers are optimistic.
India Today
12-06-2025
- India Today
First-of-its-kind brain computer helps man with ALS speak in real-time
In what could be one of the bioggest breakthrough in medical science and technology a newly developed investigational brain-computer interface could restore voice of people who have lost the team from University of California, Davis succesfully demonstrated this new technology, which can instantaneously translate brain activity into voice as a person tries to speak. The technology promises to create an artificial vocal details, published in journal Nature, highlight how the study participant, who has amyotrophic lateral sclerosis (ALS), spoke through a computer with his family in real time. The technology changed his intonation and 'sang' simple melodies. 'Translating neural activity into text, which is how our previous speech brain-computer interface works, is akin to text messaging. It's a big improvement compared to standard assistive technologies, but it still leads to delayed conversation. By comparison, this new real-time voice synthesis is more like a voice call,' said Sergey Stavisky, senior author of the investigational brain-computer interface (BCI) was used during the BrainGate2 clinical trial at UC Davis Health. It consists of four microelectrode arrays surgically implanted into the region of the brain responsible for producing speech. The researchers collected data while the participant was asked to try to speak sentences shown to him on a computer screen. (Photo: UCD) advertisement'The main barrier to synthesizing voice in real-time was not knowing exactly when and how the person with speech loss is trying to speak. Our algorithms map neural activity to intended sounds at each moment of time. This makes it possible to synthesize nuances in speech and give the participant control over the cadence of his BCI-voice,' Maitreyee Wairagkar, first author of the study system translated the participant's neural signals into audible speech played through a speaker very quickly — one-fortieth of a attributed the short delay to the same delay as a person experiences when they speak and hear the sound of their own technology also allowed the participant to say new words (words not already known to the system) and to make interjections. He was able to modulate the intonation of his generated computer voice to ask a question or emphasize specific words in a process of instantaneously translating brain activity into synthesized speech is helped by advanced artificial intelligence researchers note that "although the findings are promising, brain-to-voice neuroprostheses remain in an early phase. A key limitation is that the research was performed with a single participant with ALS. It will be crucial to replicate these results with more participants."
