You are on page 1of 4

Deep Learning in Health Sector of Bangladesh

The current pandemic's outbreak has caused enormous changes in the healthcare industry around
the globe. In order to prevent extra inconveniences and the possibility of contamination, both
doctors and patients are favoring online services. Electronic Health Records (EHR) are
consequently becoming very flooded. The text-based records must be sorted according to
medical specialization for the system to be properly and simply maintained, and automation is
essential to attaining this goal. In actuality, with sufficient training and reliable data, both
machine learning and deep learning techniques can be very effective. The data component is
what makes the healthcare industry such a profitable one for AI. Around 50 petabytes of data are
produced annually by hospitals around the world, primarily from medical imaging. To put this in
perspective, 50 petabytes of data are equivalent to all of humankind's written works from the
dawn of recorded history in all languages. Hospitals produce this much data annually, and the
best thing is that 97% of it is wasted (Ahnaf, A., Orpa, T. H., & Rizu, M. I.,2021)
Over $4 billion was invested in AI healthcare startups in 2019 alone, and this number looks to be
rising. There are currently 90+ healthcare AI startups to watch, according to CB Insights. These
startups work in a variety of fields, including imaging and diagnostics, drug discovery,
genomics, fitness, virtual assistants, clinical trials, nutrition, compliance, mental health, remote
monitoring, hospital decision support, predictive analytics and risk scoring, and much more.

In accordance, the models may give classified medical specialization of the relevant dataset and
grasp the context of each patient in their own natural language. The Bidirectional LSTM model
that we utilized in this case to train the machine learning and deep learning models yielded an
accuracy of 82% (Wang, X., Liang, G., Zhang, Y., Blanton, H., Bessinger, Z., & Jacobs, N. ,
2020). In a world where every industry is occasionally becoming digitalized, complete machine
dependence for the healthcare sector may be debatable, but it may undoubtedly be a very positive
future.Big data, artificial intelligence, and machine learning applications have not yet started in
Bangladesh. In light of this, one can contend that it is too early to introduce legislation or that
doing so could prevent the development of such technology.

Mostly used in Healthcare Sector:

● Genomics - Deep learning techniques are used in genomics to analyze a genome and
assist patients receiving therapy in gaining knowledge of potential future illnesses that
may affect them. With the aid of the insurance industry, deep learning is expected to have
a bright future in the field of genomics, which is constantly expanding. To make medical
field researchers and doctors quicker and more accurate, deep learning techniques are
applied.
● Cell scope - A Cell scope uses deep learning technology and aids parents in continuing to
keep track of their children's health issues or health status. This technology, which is
accessible on all devices, reduces the number of times parents visit hospitals or medical
facilities to visit their kids or to see a doctor.
● Medical imaging - Several methods are employed in the medical field to diagnose
hazardous disorders, including CT scans, ECGs, MRIs, and others. Heart attacks, cancer,
brain tumors, and other serious illnesses are among the dangerous conditions. As a result,
deep learning can be used to consult medical professionals who can accurately diagnose
and treat patients.
● Drug discovery - Data learning is used to both discover new medicines and to develop
them. An analysis of a patient's medical history is performed, and therapy
recommendations are made as a result. We can learn from the patient's tests, reports, and
information about the disease's symptoms thanks to deep learning in healthcare.
● Fertility rate - Over the past few decades, nations have adopted a number of excellent
approaches to enhance results in the areas of nutrition, reproductive, maternal, neonatal,
and child health (RMNCH-N). According to a recent worldwide review, nations that
focus on a few context- and time-specific RMNCH-N service factors can lower the
fertility rate in several Asian contexts to 2.5 in comparison to generic health
interventions. However, because of their complexity and fluctuation, determining the
importance and size of demand and supply-side variables is challenging and necessitates
constant monitoring. An effort was made in a recent policy research working paper to
address this issue by identifying priority investments that could help Bangladesh advance
toward RMNCH-N usage using Machine Learning (ML) techniques
(Technostacks,2019).
In order to meet their objectives, the health sector is one such platform that uses technology like
deep learning. Medical professionals look for different approaches to incorporating new
technologies. For a better future, these technologies must have an impactful effect. Deep learning
compiles a large amount of data, including patient records, medical reports, personal
information, insurance reports, etc. to give patients better care for a successful conclusion.

Limitations:

Before these technologies are widely used, there are still a lot of issues to think about and a lot of
laws, policies, and frameworks to be passed. But one thing is certain: new technology has the
potential to revolutionize how patients, physicians, and administrators interact with one another.
Perhaps it should be seen as a resource for patients, clinicians, scientists, administrators, and
everyone else working in the healthcare business rather than as a replacement for doctors.
Perhaps it is time for Bangladesh's healthcare administrators and doctors to start considering the
potential of AI — not as a replacement, but as an assistant, in a nation where the doctor-patient
ratio is 5.26 per 10,000 people (the second worst in South Asia), according to WHO, and
healthcare inequity is severe (Technostacks,2019). We might just need an extra helping hand
from technology to make healthcare available to everyone.

In terms of learning ability, pervasiveness in medical contact, making suggestions, and


opaqueness of producing such recommendations, artificial intelligence-based medical devices or
software differ from traditional medical software. Such developing technology cannot be
regulated by the current medical device regulations. As a result, Bangladesh's current legal
frameworks for medical devices are insufficient to cover artificial intelligence-based devices,
creating regulatory uncertainty over the use of these devices. To ensure the safe and efficient use
of medical devices based on artificial intelligence in Bangladesh, the appropriate regulatory
authorities should think about implementing a new policy framework that adheres to current
international norms. Bangladesh regulators should consult the recent FDA discussion paper
(FDA, 2019) and the framework, principles and guidance of the IMDRF to establish sustainable
and harmonized approaches for regulatory control of artificial intelligence-based medical devices
(Technostacks,2019).
Future:

Insights into deep learning in health care future applications are evident and can be clearly
observed. Some of them are:
➢ The latest technology of deep learning is coming to provide new and efficient paradigms
in obtaining the end to end learning models for complex data types.
➢ Usage of electronic health records (EHR) promises to advance clinical research and better
inform the decision-making skills clinically. Modern electronic health records can
prevent the practice of predictive modeling by summarizing and representation. Patients
who achieved results based on electronic health record data and an alternative feature
learning strategy are being successful.
➢ Electronic health recorder or EHR with predictive modeling Dara is anticipated to solve
personalized health quality in medicine. Data learning approaches can be used to create
an accurate and scalable prediction for a variety of sceneries regarding clinical resources.
In deep learning, neural networks can be used for identifying relevant information from
the patient’s chart and records.
➢ Deep learning can further be used in medical classification, segmentation, registration,
and various other tasks. Deep learning is a steadily developing trend in the field of data
analysis and has also been named one of the 10 breakthrough technologies
(Technostacks,2019).
➢ Deep learning is an advancement of artificial neural networks which consist of more
layers at higher levels of abstraction. There is an improvement in the predictions from the
data developed using deep learning algorithms. Deep learning is emerging as a very
important machine learning tool in imaging, convolutional neural networks, computer
domain vision etc.

References

1. Ahnaf, A., Orpa, T. H., & Rizu, M. I. (2021, November). Medical Specialty
Classification from A Bangla Dataset: A Token Level Approach Using Several Machine
And Deep Learning Algorithms. In 2021 5th International Conference on Electrical
Engineering and Information & Communication Technology (ICEEICT) (pp. 1-5). IEEE.
2. Wang, X., Liang, G., Zhang, Y., Blanton, H., Bessinger, Z., & Jacobs, N. (2020).
Inconsistent performance of deep learning models on mammogram classification. Journal
of the American College of Radiology, 17(6), 796-803.
3. Technostacks. (2019, January 4). Deep learning in healthcare - Benefits - Current &
future usages. Technostacks Infotech. https://technostacks.com/blog/deep-learning-in-
healthcare/

You might also like