Technology For People: Health
The current pandemic has shown that our healthcare systems are not yet up to the mark. But what can we definitely conclude is, we have come leaps ahead as to how the previous ones were handled. That being said, we definitely need to improve our systems. One of the ways or methods that can be used to improve is precision medicine. Precision medicine is an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment, and lifestyle for each person. This will enhance the ability of doctors and researchers for the prevention and treatment of a particular group of people together more effectively. Image analysis and natural language processing are going to play a key role in the success of Precision Medicine.
The systematic analysis of medical images to extract information by segmentation, thresholding an application of various machine learning techniques can be used for image analysis. Natural language processing (NLP) is part of artificial intelligence, specifically human interface intelligence, wherein, computational techniques such as recurrent neural networks, hidden Markov models, generative models such as Viterbi are employed to synthesize and semantically parse natural language and speech. It could help in understanding written prescriptions, reports, and help create a final result.
Image Analysis
There are several methods used to capture medical image data, some of the key ones are noted below:
- X-rays
- Computational tomography (CT) or computed axial tomography (CAT)
- Magnetic resonance imaging (MRI)
- Nuclear imaging modalities such as positron emission tomography (PET) and single-photon emission computed tomography (SPECT)
The advancements in CT, MRI, and PET image scanning have enabled the creation and storage of high-quality data in the shorter time frame as compared to X-Ray images which have been abundantly used for research and advancements. Images created from various modalities are stored in information systems such as picture archiving and communication systems (PACS) and the widely used format to store, exchange and transmit such images is DICOM (Digital Imaging and Communications in Medicine).
The advancements of image analysis can help in pre-processing images such as to understand the extent of tumors and whether chemotherapy can be applied or not. The advancements in techniques using Deep Learning Networks, Genetic algorithms, and state of art GPU’s have enabled such processes.
Natural Language Processing
Most of the healthcare data is available in free text form and information extracted from them can be used to improve the treatment and prevention of disease for people of a particular area or genetic background. For example, NLP methods can extract specific mentions and details of a radiotherapy treatment from a clinical note, or emotional concerns in a patient’s discharge summary.
Application of Image Analysis and NLP Together
Both of the technologies work complementary to each other as NLP can extract information on reports and image analysis can do an in-depth analysis of image reports. Both of them can combine to give qualitative results.
Challenges
There are a couple of major challenges in the usage of both technologies. The image data is of gigantic size and the creation of a copy of such images is harder. For NLP based analysis, multi-lingual data creates a huge problem. Both of them combined have the issue of privacy which is of major concern for all the individuals.
Opportunities
There are great opportunities ahead when the application of image processing and NLP techniques mature and the said challenges are overcome. Together, these techniques can greatly improve the productivity of the clinicians while assisting in their work with valuable insights that otherwise would take great amounts of time to populate. For example, in the case of oncology, the image analysis can not only assess how well a treatment is working but also compare with millions of similar images and predict an optimal pathway that worked best in other similar patient populations. At the same time, NLP can be employed to corroborate information from other sources such as pathology reports, lab reports, radiology reports, and physicians’ notes and thereby auto-generate a clinical report using standard nomenclature and terminology and by assessing precise recommendations on further treatment.
Both image analysis and NLP can also be applied to large amounts of medical image data and associated notes to create clusters of similar observations that can enable other initiatives such as stratification of patient populations and population health initiatives.
Ethics
Ethics plays a major role in healthcare data as privacy and confidentiality should be maintained at any cost. The
- Assurance of respect for all participating subjects that participating individuals will be treated as autonomous agents and individuals with diminished autonomy are entitled to additional protection
- Ensure voluntary participation of individuals as research subjects with secured informed consent without any undue influence and/or coercion
- Ensure beneficence. Two general rules to assure beneficence are: (a) do not harm and (b) maximize possible benefits and minimize possible harms
- Fairness in patients selection and assurance of social justice in the selection of patients
- Safeguarding the privacy and confidentiality of the protected health information captured
Conclusions
Both of these technologies are maturing with time. As these technologies advance, we will have better and far more accurate and faster results in genome sequencing, editing, and precision medicine.
