Redacción Farmacosalud.com
The XXIX National Congress of the Spanish Society for Bone Research and Mineral Metabolism (SEIOMM), a meeting recently held in Mallorca, was the setting in which a novel screening algorithm was unveiled that uses data from medical records processed by AI (Artificial Intelligence) to detect possible cases of X-Linked Hypophosphatemia (XLH). The new methodology, which still needs to be perfected, has yielded encouraging results, since it has an accuracy of 61% when categorizing patients at high risk of suffering from XLH, once compared with the opinions of experts.
From a reference population of 1.2 M patients, and after applying the corresponding inclusion and exclusion criteria, 58,034 individuals from five hospitals were preselected. The most frequent criteria presented by these people were joint pain and muscle fatigue and, at a biochemical level, altered kidney function. The algorithm was applied to this preselected population, within the framework of a pilot study.
Dr. Pedro Arango Sancho
Source: Omnicom PR Group (FILE IMAGE)
The tool is only applicable to the adult population, but work is already underway on a pediatric adaptation
“The most important thing was that, when both systems were compared (the algorithm designed in this study and the opinion of the experts), we saw that the experience was quite good for a first approximation. We agreed with the new methodology in the categorization of 33 of the 54 patients classified as high risk. This means, in data, that the algorithm had an accuracy of approximately 61% for those at high risk, around 30-35% for those at medium risk and around 70% for those at low risk,” specifies Dr. Pedro Arango, pediatrician at the Nephrology Unit of the Sant Joan de Déu Barcelona Hospital.
“This gives us an idea: that of having carried out a pilot study that lays the foundations to continue strengthening and refining the algorithm. We know that there are still certain limitations, such as the fact that the current design of the tool makes it applicable only in the adult population. However, we are already working on another algorithm adapted to the pediatric population in order to detect cases that may have gone unnoticed. The truth is, we are very happy with this new advanced methodology because we believe that it can become a very, very useful tool in the future,” he says. Arango.
“The algorithm aims to facilitate initial screening for the clinician”
“AI has been established as a strategy or as a measure to carry out this project due to the fact that its use allows the method to be standardized and extrapolated, which in turn will make it possible to ensure that XLH is not a disease whose detection depends on expertise* of who values it or who thinks about it, as happens today. With all this, the aim is for the pathology to be diagnosed to a greater extent, given that the algorithm aims to facilitate initial screening for the clinician, thus reducing the possibility of potential cases being omitted. In this way, cases of possible risk could be identified so that, later, the doctor could further evaluate the indicated patient. In other words, the algorithm would act as a support tool for the clinician,” explains the specialist.
*expertise: special skill or knowledge
The new system uses a technology called Natural Language Processing (NLP), the application of which allows information to be extracted from computerized medical records. On this occasion, it was used in five Spanish hospitals to then, based on the data obtained, establish risk categories: high, intermediate or low risk (according to the combination of factors that have been established to have hypophosphatemia mediated by FGF23). This included not only XLH, but also another disease: tumor-induced osteomalacia (TIO).
During the pilot test, a subgroup of anonymized data was configured and shared with five clinical experts in the field in order to validate the algorithm. “In a completely anonymous and blind manner, that is, without knowing what the risk algorithm had determined for these patients, each of us developed our own classification based on the available information. Then, the risk determined by the algorithm was compared with the opinion that we, the experts, had,” comments Dr. Arango.
Autor/a: Maximusdn
Fuente: depositphotos.com
Regarding the variables used, they were mainly clinical. With them, the aim was to evaluate whether the subject presented any feature of the typical symptoms of this type of hypophosphatemia. Biochemical variables were also considered, which were obtained from blood tests, including parameters such as phosphate. Based on the combination of the different indicators, the algorithm made the three risk determinations and, subsequently, the results were compared with the physicians’ clinical classification.
Skeletal deformities, bowing of the legs…
XLH is a rare condition that causes phosphate deficiency and manifests itself in childhood. These pediatric patients suffer from skeletal deformities, bowed legs, cranial anomalies, motor delay, short stature, muscle weakness, and bone and joint pain. Early detection is key, since the low prevalence of XLH and the limited clinical knowledge about the pathology are factors that sometimes contribute to underdiagnosis.
Another serious setback is the late diagnosis of XLH, since “it can have many repercussions, especially from a functional point of view. However, it is true that we take into account patients suffering from the classic phenotype, who are those whose appearance is very characteristic of XLH, hence these patients do not usually go unnoticed. In fact, and although we have seen some more extreme cases, they are conditions that do not usually go unnoticed,” says the doctor.
“But those patients who present less associated physical affectation (less physical stigma) could go unnoticed more. If so, if they were not detected, the quality of the bone could not be improved during a determining period such as childhood. They would reach the end of the pediatric stage, in adulthood, with a bone in very poor condition. This could condition the appearance of joint and muscle pain, problems with repeated fractures, decreased quality of life and functional decline. All of this greatly affects the day by day of the sick,” he warns.
Autor/a: Syda_Productions
Fuente: depositphotos.com
The diagnosis of XLH, made difficult by the genotype/phenotype variability of the disease
XLH, This is an alteration that means that the kidney and intestine are not able to adequately absorb phosphate. “This phosphate that is lost through the urine and in the intestine in an increased manner hinders the correct formation of bone and causes the bones of growing children to arch, giving rise to rickets. In adults it generates poor bone quality that ends up causing what we call osteomalacia,” details Arango.
The genotype/phenotype variability of the disease is another problem that makes the diagnosis of XLH difficult. This means that it is a pathology in which the genetic alteration will not always determine the severity of the condition. Thus, it may happen that patients who have the same mutation in the same site of the gene that causes the disease, the PHEX gene, present completely different effects in terms of severity, with some being very mild and others very severe.
As a consequence of the aforementioned variability, very severe patients can be seen who debut with the characteristic phenotype, which are bowed legs or widening of the wrists in the metaphyseal area, but also others who do not really present so many external bone characteristics and whose involvement in adulthood is sometimes confused with osteoporosis or fibromyalgia, because they suffer from “chronic functional pain that limits them in their daily activity,” says the pediatrician of the Nephrology Unit. of the Sant Joan de Déu Barcelona Hospital.
