Fuzzy Lab - How Artificial Neural Networks Can Help Medicine

Some researchers of the Saint Raphael Institute of Milan have
discovered a useful software for the analysis of proteins and very
significant in the diagnosis of some pathologies of the immune system.
The new program, known as FuzzyLab, concurs to obtain better results
and to pull down of 80% the times of examination, to all advantage of
the doctor but also of the patient. FuzzyLab is a software
neuro-fuzzy, it means that is based on a combination of neural
networks and fuzzy logic.

The neural networks are similar to an artificial brain in a position
to continuously improving the own ability to learning. Fuzzy logic
renders the processes of calculation more flexible. This type of logic
is applied already from some year to a lot of products of wide
consumption for instance it provides the washing machines of last
generation to measure out water and detergent based on the level of
stains in laundry. Fuzzylab carries out quickly all the automated job
of interpretation of the analyses until today executed manually and
entrusted to the interpretation of the doctor.

"In a laboratory of great dimensions" - explains Stefania Del Rosso,
the responsible of the Laboraf-Research laboratory of Saint Raphael
Institute- "350 analyses are made every day in average, whose reading
demands to an expert doctor at least two hours of job. to do the same
task the software employs only 20 minutes. In this manner the doctor
can distinguish quickly between regular "and pathological outcomes"
and dedicate so the saved time to the investigation of examinations
more hard to understand ".
The capillary analysis examines proteins of the serum and, beyond
providing general indications about inflammatory states and possible
hepatic suffering, it is able to reveal the presence of anomalies in
the immune system. Is the situation for instance of the monoclonal
members - a group of equal antibodies whose onset is correlated to
malignant lymphatic pathologies like the skin tumour, the chronic
lymphatic leukaemia and the lymphomas.
to learn to read the analyses, FuzzyLab has been subordinate to an
difficult training articulated in three moments. In the first phase
the software has been instructed to recognize the shape of the regular
analytical curves through a mathematical algorithm, on the pedestal of
the data supplied from the doctor "trainer". In the second phase have
been inserted in the computer the mathematical formulas for the
identification of the pathological curves. In the third phase
(validation of the system) the manual interpretation and that among
FuzzyLab have been put to comparison to estimate of the error margin,
resulted inferior to 2%.
The system is extremely flexible and potentially in a position to
increasing just the modernization with the experience. "The advantages
of this approach are significant both for the doctor and for the
patient", says Michelangelo Murone, assistant manager of Laboraf. "The
most clear one is the reduction of the times of reporting that
provides the analyst to have more time disposition to examine the
pathological tracings. The second advantage is the improvement of the
interpretative quality, translated in a surer reporting, because the
pathological curves are analyzed at least two times, before from the
software and so from the doctor, but also more objective and
standardized, because Fuzzylab reduces the probability of mistakes and
deletes the risk of interpretative differences regarding to the human
factor "