Title : New probabilistic diagnostic aid tool for cardiac dynamics
Abstract:
Background: Probability theory and dynamical systems theory have been applied to cardiac dynamics, leading to the development of new methodologies that differentiate patients with normal medical diagnoses from those with chronic and acute conditions.
Materials: A total of 120 Holter monitor recordings, each lasting 21 hours, were analysed from patients aged over 21. Three prototype cases were selected, corresponding to normal, chronic, and acute diagnoses. An induction process was carried out to develop a general probability space using repetition ranges of 1001–2000 and 2001–3000 in heart rate data. The probability for each range was calculated, and the results of the most frequent probabilities were summed.
Method: A blind study was conducted using the remaining Holter records. Diagnostic outcomes were concealed, but cardiac dynamics were made available. The same methodology used on the prototypes was applied to the blind study data. The probability results from the blind study were then compared with those from the prototypes, and their agreement was confirmed through statistical analysis, including sensitivity, specificity, and kappa coefficient.
Results: In normal dynamics, repetition ranges from 1001 to 2000 varied between 14 and 11, and from 2001 to 3000 between 3 and 0. In chronic dynamics, repetitions ranged from 31 to 21 (1001–2000) and from 14 to 10 (2001–3000). In acute dynamics, the 1001–2000 range varied between 11 and 9, and the 2001–3000 range from 6 to 3. The probability loads for normal dynamics in the 1001–2000 range were between 0.46 and 0.35, and in the 2001–3000 range between 1 and 0. For chronic cases, probabilities in the 1001–2000 range were between 0.48 and 0.35, and between 0.7 and 0.54 in the 2001–3000 range. For acute cases, the 1001–2000 range showed probabilities between 0.6 and 0.5, and the 2001–3000 range between 0.75 and 0.46. The probability loads in the 1001–2000 frequency range were between 55 and 95 for normal cases, 65 and 105 for chronic, and 75 and 100 for acute. In the 2001–3000 range, loads varied from 65 to 95 (normal), 65 to 85 (chronic), and 80 to 100 (acute). The statistical analysis showed a sensitivity, specificity, and kappa coefficient of 1.
Conclusions: A clinical diagnostic support tool was developed, capable of distinguishing between normal, chronic, and acute states based on the probabilistic load of heart rate frequencies, with clinical applicability.