Examples of DA applications in medicine

The examples considered below are described in more detail in the demo datasets supplied with DALSolution.

This problem was encountered during prophylactic ultrasonic study of kidneys among pilots of civil airlines. Devices SSD-2000 by Aloha and SSH-140AG by Toshiba were used.

During an ultrasonic study, it is difficult to visually distinguish multiple kidney sinus cysts from hydronephrosis, even for an experienced doctor. The “multiple cysts” diagnosis allows a pilot to continue flying. The “hydronephrosis” diagnosis requires the pilot to stop flying. An error means either the unjustified end of a pilot’s career, or the unacceptable risk of passengers’ lives.

Problem. To design a diagnostic rule differentiating multiple kidney sinus cysts from hydronephrosis, using kidney characteristics measured with the applied ultrasonic devices.

The problem was being solved at the Department of Ultrasonic Diagnostics, Russian Medical Academy. With the help of DALSolution, it was discovered that an important role is played by the resistance index (RI). The bounds on RI were computed such that one diagnosis could be distinguished from another. Also, accurate decision rules leading to a correct diagnosis were found. Here is an example of a rule diagnosing multiple kidney sinus cysts:

IF the kidney Resistance Index is not higher than 0.69, THEN, from two alternative hypothetical diagnoses “multiple cysts of kidney sinus” and “hydronephrosis”, one should select “multiple cysts of kidney sinus”.

In all cases observed during the clinical study, when this rule was applicable (there were 38 cases) it was confirmed 100% of the time, without error, and it explained over 80% of all cases with this diagnosis (A. N. Hitrova, 1996, S. V. Chesnokov, 1997).

A Moscow hospital conducted clinical testing of hormonal drugs used to alleviate negative menopausal conditions. Three drugs were compared, which we will call Drug 1, Drug 2, and Drug 3.

Problem. Find the rules governing the choice of a hormonotherapy drug under different anamnesis conditions.

The study was initiated by a pharmaceutical company producing therapeutic hormonal drugs. Rules found with the help of DALSolution clarified the relationship between drug effectiveness and diseases the woman had had before treatment began. As a result, the company acquired drug-specific knowledge, while the hospital could better treat patients by considering their medical history.

Of the three drugs, only two (Drug 1 and Drug 2) achieve the necessary effect. In the following four situations, hormonotherapy resulted in an improvement or normalization of all symptoms in 100% of cases:

Situation	Drug	Condition at the time of first visit	Gynecological disease in anamnesis
1	Drug 1	Post-menopause	Myomas
2	Drug 2	Post-menopause	Inflammations
3	Drug 2	Post-menopause	Cysts in ovary
4	Drug 2	Post-menopause	Chronic salpingoophoritis

The clinical testing of bone regeneration techniques during paradontosis was conducted in one of Moscow's dentistry hospitals. Three methods were compared, which we will call Method 1, Method 2, and Method 3.

Problem. Find rules for selecting a method of bone regeneration during paradontosis based on the causes of the disease. The criterion of success is complete regeneration of the affected bone.

With the help of DALSolution, the problem was solved. Here are examples of situations when a treatment resulted in complete bone regeneration with 100% accuracy:

Situation	Causes of paradontosis	Regeneration method to be selected
1	Complicated caries	Method 1 or Method 2
2	Perforation	Method 1
3	Supracontact	Method 3

This study of relationships between children's health and the environment was conducted in one of Russia’s cities.

Problem. Find rules explaining the bad health of a child by using the environmental conditions of his/her place of residence and the mother's working conditions.

Application of DALSolution helped solve this problem, identifying the relationship between environmental conditions and children’s health.

If a mother’s work is extremely hazardous, and the environmental conditions are catastrophic in the place of residence, then the child is sick.

The rule has accuracy of 0.84. It is confirmed in 103 cases out of 123. The rule is not universal. It describes only 26% of the cases when the child is sick (rule completeness is 0.26). Other rules explain the other cases.