World Hearing Day was on the 3rd March and highlighted the common, debilitating, and preventable issue of noise-induced hearing loss, which affects many South Africans. Noise-induced hearing loss (NIHL) is the second most common occupational disease[1] in South Africa, after tuberculosis. This is a particular challenge for industries such as mining, where heavy equipment, rock drilling, and a confined work environment all expose employees to high noise levels.

“As the name suggests, noise- induced hearing loss is caused by exposure to loud sounds,” says Prof Dr Wim Delva, founder and co-director of Wimmy, leading data science specialists in health. “The louder the noise, the shorter the period needed to cause lasting damage: Exposure to noise at 85 dBA (roughly the noise level of heavy traffic) may damage hearing if it lasts for more than eight  hours[2][3], while music played through headphones at the highest volume, can, at 110 dBA, cause damage in just minutes.”

According to Dr Delva, noise-induced hearing loss affects people of all ages, including children, teens, young adults, and older people.  A 2011-2012 CDC study notes that up to 24% of adults under the age of 70 have a degree of NIHL, while a 2005/6 study estimates that as many as 17% of teens (ages 12 to 19) may experience NIHL[4].

 This, in turn, can undermine an individual’s quality of life. Hearing loss can result in social withdrawal, misunderstandings, and a higher risk of injuries. Hearing loss is also associated with a higher incidence of dementia[5].

Companies are required by law to do annual audiograms on employees exposed to noise, to monitor for possible hearing loss. Applying data science and analytics can make this process far more effective and objective, and go a long way to prevent this irreversible disease.

Keeping an ear to the ground

“Using data science and analytics can ensure that risks are identified and quantified, and that early signs of NIHL are picked up and actioned,” says Dr Delva.

 “False diagnoses can also be minimised, and more serious cases prioritised. It also provides a way to evaluate how effective existing controls are, and to identify areas where they are failing. Evidence-based recommendations can then be used to improve existing programmes – for example, we create automated reports and dashboards that give companies objective, shareable insights into the hearing health of the organisation, so that they’re in a better position to act.”

He provides several examples of ways in which data science is being used to help companies make their hearing conservation programmes more effective.

The potential value of AI

One of these examples involves analysing the tens of thousands of audiograms carried out by a company over many years, to determine whether using Artificial Intelligence (AI) and diagnostic algorithms can significantly improve the speed and accuracy of the interpretation of tests, especially when used alongside clinicians.

 This matters, because interpreting audiograms requires highly skilled audiologists – a scarce resource – and the interpretation of audiograms and recognition of early signs of NIHL can vary greatly.

 Depending on how accurate the automated interpretation of tests is, this could serve as an initial screening tool, to cut down the volume of work. This could also serve as a form of quality control, reducing human error that leads to  false positives and false negatives. Lastly, if automated interpretation proves to be more accurate than human practitioners, it could be used to do all the interpretation of audiograms.

Early detection, better prevention

Within the mining industry, data analytics can identify at-risk individuals and detect those showing some degree of NIHL, early on. In this way, severe hearing loss may be prevented. It can also help to identify underlying factors that may make some employees more susceptible to noise damage, along with work areas where employees are experiencing more significant hearing changes – and why this is happening. This enables intervention before people suffer incapacity and disability.

Investigating emerging technologies

At present, screening pure tone air-conduction audiometry is commonly used to conduct hearing tests in the work environment. However, this incorporates subjectivity from those being tested, and there is a high degree of intra-, and inter-operator variation in interpreting test results. Data science can play a useful role in investigating emerging technologies for more objective options that can also handle large volumes of testing.

Hear for life

“Companies go to great lengths to avoid noise-induced hearing loss, but there is clearly room for improvement, given its prevalence and the seriousness of its consequences,” says Dr Delva. “At Wimmy, we are developing automated tools to better protect employees’ wellbeing, and safeguard their hearing, for life.”

[1] What Is The Most Commonly Reported Type Of Occupational Illness? – Activekyds

[2] https://www.workplacetesting.com/definition/4081/decibel-a-scale-dba

[3] https://www.nidcd.nih.gov/health/noise-induced-hearing-loss

[4] Prevalence of Noise-Induced Hearing-Threshold Shifts and Hearing Loss Among US Youths – Elisabeth Henderson, BA; Marcia A. Testa, MPH, PhD; Christopher Hartnick, MD, MS

[5] https://www.nia.nih.gov/health/hearing-loss-common-problem-older-adults