INFORMATICS
AI , Automation and Public Health Laboratories
By Rebecca Abelman , senior specialist , AMR Informatics
The use of automation and artificial intelligence ( AI ) is growing in clinical laboratories , and new instruments and solutions are debuted regularly . These innovations aim to decrease bench time , increase test sensitivity / specificity , and decrease turnaround time for results . These new technologies will shape the future of public health laboratories and data .
“ Artificial intelligence ” is a broad term used to describe computer systems or algorithms used to imitate human intelligence . There are various types of AI and various ways it can be applied , but the most relevant to healthcare is machine learning . While traditional programing requires a human programmer inputting predefined patterns into an algorithm to create rules for analyzing data , machine learning instead uses a computer model to develop an algorithm that creates the analysis rules . Traditional programming can create an algorithm that is defined and customizable by the programmer , but machine learning algorithms can be created for large input variables and complex data that exceed the limits of human capacity . This ability to handle larger , multiplex data sets allows for instrumentation that perform tasks typically performed by humans and can even out-perform them in some cases .
Automation is already implemented across a variety of laboratory technologies and it is a vital part of modernization of laboratories around the world . Laboratory automation is the utilization of technologies that can be controlled with minimal scientist interaction . Most automated instruments in laboratories today are classified as partial automation , as they automate one process in a testing workflow . This includes all-in-one PCR instruments and culture systems . However , full workflow automation instrumentations are entering the market , and these instruments provide beginning-to-end automation of entire assays . Some even include sample processing so laboratorians can load samples directly into the instrument . These workflow automation instruments increase walkaway time for scientists , decrease error and turnaround time , and facilitate interoperability between different laboratory technologies and systems .
A Place in the Public Health Laboratory
Practical applications of AI and automation are already out in the field . Digital microscopy image analysis programs for hematology and microbiology can review thousands of images using AI and learn to alert laboratorians of abnormal / significant view fields or pathological conditions present in a sample . This AI type is found in the Beckman Coulter Scopio™ System , which uses an AI-based application to view peripheral blood smears and claims to reduce turnaround times by 60 percent . The BD Kiestra™ is an example of both AI and automation used to create a collection of workflow automation instruments for microbiological culture . The Kiestra™ system uses incubators attached to a series of cameras to create digital images of culture plates , and some models include culture plate tracks to deliver plates from the specimen processing units to incubators to connected workbenches . Screening culture plates like Methicillin-resistant Staphylococcus aureus ( MRSA ) and Urine plates can be read by AI applications trained to interpret negative or positive cultures by colony morphology .
The impact of AI and automation on public health is still being explored , and many of these innovations could have a major impact in the field of public health . The high data and variable capacity of AI analysis could be used by epidemiologists , research scientists and public health laboratories for improved data analysis . Increased automation could decrease the burden on public health laboratorians by decreasing bench time . These technologies could also create new data elements that need to be added to databases — a process that takes extensive time and resources . The scope and scale of these improvements are dependent on how accessible these technologies are and whether they produce quality data and test results . Even if these tools are implemented , the current lack of standards and quality assurance resources needs to be addressed before these solutions can be utilized in public health laboratories . Preparation for the current and eventual application of more AI and automation in public health is crucial for supporting the future of public health laboratories , scientists and data . g
30 LAB MATTERS Fall 2024
PublicHealthLabs @ APHL APHL . org