Donated blood has been evaluated by medical personnel for decades by using conservative and traditional methods. However, an international study that has made use of Machine Learning may soon change this conventional evaluation method.
A team of researchers has brought forward the use of Machine learning to analyze the shape of red blood cells present in the stored samples. This new application of machine learning can be a huge development in the Precision Diagnostics Market. It can provide medical personnel with the ability to do blood analysis faster while also being precise. This may bring changes in how donated blood is assessed, stored, and selected for transfusion.
Once blood is taken outside the body and stored, red blood cells start changing their shape as they become older. Thus, leading to changes in their ability to function and carry oxygen inside body tissues at the time of transfusion. The red cell products can be stored for a maximum of 42 days, and hence, they require strict monitoring. In the present time, donated blood is evaluated through a glass slide, wherein a drop is placed, and the cells present are then analysed. This method helps classify the blood cells by judging the shapes of cells present in a sample of 100. Then, a morphology index is calculated (score is given based on the shape of sample cells).
This process turns out to be very time-consuming and looking at a small sample and evaluating the blood also makes it subjective. To make the process faster and more accurate, the researchers used the potential of Artificial Intelligence.
The team used imaging flow cytometry to help their cause. It is a technology that captures images of thousands of cells from a single drop of blood. And then creates a large database to conduct analysis. These images contributed towards automating the traditional expert assessment by training the computer. The automated process worked on assessing more than 100 blood samples. A task that takes months for technicians to complete can now be completed in a day.
Researchers also found a solution to making the evaluation less subjective. They commanded the AI to look at different parameters, and were able to pick up subtle differences that evaded from the human eye till now. The team stated that red cells don’t go from one shape to another but gradually progress to a different shape. With the help of AI now, we can better classify these subtle differences.
The study's conclusion has equipped the research team with algorithms that can present the usefulness of machine learning in qualifying red blood cells precisely. Hence, practitioners can now be more accurate in their ability to match blood donors and recipients of the blood based on their blood characteristics.