Predictive Medicine Powered by Artificial Intelligence
Using advances in technology to support clinical decision making
By combining in-depth expertise in medical biology with advanced computer technology, Biron offers health care professionals clinical decision support aimed at predicting health problems that are often unsuspected. Based on routine blood tests, these predictions will assist them in considering complementary analyses if they are deemed clinically relevant.
Did you know?
Innovations in artificial intelligence are now being taken very seriously or are already in use among various medical specializations. In addition to radiology and pathology, these include oncology, dermatology, ophthalmology and much more. [1] While this list is quickly growing, what is happening with laboratory medicine?
A publication from Harvard Medical School demonstrated that it is possible to accurately differentiate between normal and abnormal ferritin levels without directly measuring them. This achievement was possible thanks to a mathematical and statistical approach whereby a computer processes an enormous amount of information, including demographic, biochemical and hematological data from patients. It can also improve its own problem-solving performance (also known as machine learning). In addition, the study found that ferritin levels predicted by artificial intelligence may in certain cases better reflect the patient’s actual iron status than by measuring ferritin directly.
Prediction process
4 possible scenarios:
- Ferritin level is predicted to be abnormally low when iron deficiency is not suspected. Consider a ferritin test if clinically relevant.
- Ferritin level is predicted to be abnormally high when iron overload is not suspected. Consider a ferritin test if clinically relevant.
- Ferritin level is predicted to be significantly different from the measured ferritin. e.g. normalization of ferritin levels due to a mixed status (inflammation masks an underlying iron deficiency).Consider prediction if clinically relevant.
- No abnormalities are predicted.
2 possible scenarios:
- Identification of an elevated glycated hemoglobin (greater than or equal to 5.5%) when a metabolic disorder is not suspected.
- No abnormalities are predicted.
2 possible scenarios:
- Identification of an elevated PTH, while hyperparathyroidism is not suspected.
- No abnormalities are predicted.
What should be done in case of an abnormal ferritin prediction?
- Biron may analyze the patient’s ferritin, free of charge*, based on the sample used for the initial blood test.
Contact Biron - Direct line for health care professionals: 1-866-923-9222 ext. 2846 - Prescribe a ferritin test for the patient, with a new sample.
*Subject to change
- Prescribe the dosage of PTH or HbA1c to the patient from a new sample, free of charge*.
*Subject to change
Important considerations for health care professionals
- Under no circumstances will the measurement of the abnormal predicted analyte be performed as a reflex. The decision to act on a prediction is entirely up to the clinician.
- Patients do not have access to the predictions.
- Predicting ferritin is just the beginning for Biron. More algorithms for the early detection of health problems will follow.
Sources1
- CADTH. An Overview of Clinical Applications of Artificial Intelligence. CADTH, Ottawa; 2018 (CADTH issues in emerging health technologies; issue 174).
FAQ
Where does the data used by the algorithm come from?
The data used are data observed in the laboratory that have been anonymized before being used by the algorithm.
How long is the data used retained?
We retain data for as long as it is needed to provide our services, to manage our business, and to comply with our legal and regulatory obligations (e.g., health care providers have statutory retention obligations with respect to patient records). When we no longer need it, we securely destroy the data or anonymize it (so that it no longer identifies a patient).
Do you have the right to use your patients' medical information?
In accordance with our privacy policy, which our patients adhere to by doing business with us, we use personal information to provide our services, to manage our business and for other purposes permitted or required by law.
We may use personal information to conduct data analysis and research to improve or develop health-related products and services. In this case, we will generally aggregate and/or anonymize personal information so that it no longer includes information from which a patient can be identified.
For example, we may conduct data analysis and research for the following purposes at any time in compliance with applicable law:
- Determine the effectiveness of a treatment and find ways to improve it, or develop tools to improve diagnostics or imaging readings.
- Conduct health and genetic research.
- Develop new health products and services that meet the needs of our clients.
- Perform data analysis and research and development to develop new tools such as algorithms or reports.
How do you ensure that this data is secure?
Biron adheres to high security standards when it comes to the storage and access of its data.
Do you sell this data?
No, Biron Health Group does not sell any data to third parties.
Did you ask your patients for consent before using their medical data?
The use of Biron services is governed by the privacy policy of Biron Health Group and its subsidiaries. By consenting to use our services, patients consent to this policy.
By using our services, patients consent to the collection, use and disclosure of their personal information as described in the policy. In some cases, their consent may be "implied", meaning that their permission is deemed to be given based on their action or inaction at the time of collection, use or disclosure of their personal information.
If they choose not to provide us with certain personal information or if they withdraw their consent, where consent can be withdrawn, we may not be able to offer them our services.
If a Biron patient does not want their medical data to be used in yours algorithms, how can they ensure that it is not?
It is not possible to remove anonymized patient data currently used in the algorithms database.
If for a future service at Biron a patient chooses not to give us certain personal information or withdraws their consent, in cases where it is possible to withdraw consent, we may not be able to provide our services to them.
Will the patient have access to this predictive data?
No. Patients will not have access to the predictive data, which will only be included in the results report given to the referring physician and to the physicians in copy.
Will the predictive data section appear on all Biron reports?
The predictive data section will be part of the Biron result report (physician version) only if a prediction of abnormal results is present for the patient.
Why did you choose ferritin?
We consider that ferritin prediction can be particularly relevant to detect iron deficiency without anemia, unsuspected hemochromatosis or a mixed state in case of inflammation and iron deficiency.
Why did you choose HbA1c?
Prediction of HbA1c may be particularly relevant in identifying elevated HbA1c without apparent clinical signs and symptoms. This prediction can then help to detect the onset of diabetes and spare the patient from complications related to this condition.
Why did you choose PTH?
Prediction of PTH may be particularly relevant in identifying elevated PTH, while hyperparathyroidism is not suspected. This prediction may allow for early management of the condition and thus avoid complications.
Do you plan to release other predictive data?
Yes, work is underway to predict other unsuspected metabolic and hormonal disorders.