The previous discussion question identified several valuable elements for a patient database. They include information on demographics such as race, age and gender, medication, allergy, lab results and imaging studies.
Some of these elements are predefined and have predefined formats, or can be easily classified into codes or categories. Others are not predefined or need a more complicated analysis before they are categorized.
In a database of patients, structured data may include information such as medical history, laboratory results, allergic reactions, or allergies. They are collected with standardized questionnaires or forms, which can easily be coded.
In a patient’s database, unstructured data may include imaging studies and clinical notes. They are usually recorded as free text, making them difficult to categorize and analyze. Natural language processing and machine-learning techniques are making it easier to extract information from data that is not structured.
Conclusion: Both structured data and unstructured are useful in a patient’s database. Unstructured data can be analyzed and coded more easily than structured data.
References:
Chen, C., & Li, Y. (2021). Natural Language Processing and Electronic Health Record Analysis. Artificial Intelligence in Healthcare. 73-96). Springer, Cham.
Topaz, M., Lai, K., Dhopeshwarkar, N. V., & Goss, F. (2016). The need for structured data elements, and the use of coded data to support clinical research and care is increasing. Study in Health Technology and Information Informatics 225: 690-692.