Research shows that changes in voice can reveal early signs of a rare type of cancer, transforming the diagnosis

A new scientific study has revealed that subtle changes in an individual’s voice may indicate the presence of a rare type of cancer, even before current invasive methods can detect it. The discovery, driven by machine learning algorithms, promises a revolution in the early diagnosis of cancerous lesions on the vocal folds, offering a more comfortable and accessible alternative to patients.
The joint work of researchers at Oregon Health and Science University and Portland State University suggests that voice analysis, focusing on a specific marker, could become a valuable tool. Currently, diagnosing this type of cancer, known as laryngeal cancer, involves invasive procedures such as video nasal endoscopy and biopsies, which are uncomfortable and can delay the start of treatment.
The technology that deciphers the secrets of the human voice
Scientists in the United States have identified that artificial intelligence algorithms are capable of differentiating vocal patterns imperceptible to the human ear. This ability is crucial to identify the influence of cancerous lesions on the vocal folds, which alter the sound of the voice in a subtle way, but detectable by machines.
The research focuses on harmonic-to-noise ratio (HNR), an acoustic parameter that measures the energy ratio of periodic signals (harmonics) to aperiodic signals (noise) in the voice. Fluctuations in HNR, especially its variability, proved to be a promising indicator. By analyzing this data, the technology is able to draw a vocal profile that may suggest the presence of abnormalities in the vocal cords.
Significant advantages of early diagnosis by vocal analysis
Implementing a voice-based cancer detection tool would bring a number of important benefits to public health and patients. The main advantage is the ability to identify the disease in its early stages, which drastically increases the chances of successful treatment and the survival rate.
- Other notable benefits include:
- Less invasive:Replaces uncomfortable and risky procedures such as biopsies and endoscopies with simple, non-invasive voice analysis.
- Cost reduction:Potential to reduce costs associated with complex and repetitive diagnoses.
- Greater accessibility:Facilitates large-scale screening, allowing more people to be screened quickly and efficiently.
- Patient comfort:Minimizes stress and anxiety related to invasive exams.
- Agility in treatment:Early detection accelerates the initiation of therapies, improving prognoses.
Laryngeal cancer, although not the most common, has its prevalence linked to factors such as smoking and excessive alcohol consumption. Late detection often results in more aggressive treatments and lower cure rates. It is estimated that the five-year survival rate for localized laryngeal cancer can be greater than 70%, but drops considerably when the disease has already spread. A non-invasive method such as vocal analysis could change this scenario.
Next steps and challenges for clinical application
For vocal analysis to become a standard diagnostic tool, more comprehensive studies and validation in large populations are still needed. The researchers plan to expand their samples and refine the algorithms to increase the accuracy and reliability of the results. The objective is to develop a system that can be easily integrated into clinical environments, whether in doctors’ offices or even on telemedicine platforms.
In addition to initial detection, the technology can be valuable in the ongoing monitoring of at-risk patients or those who have already undergone treatment, helping to identify recurrences early. The innovation represents a significant step in the search for less aggressive and more efficient methods to combat cancer, paving the way for a new era in oncological diagnosis.






