The aggressive respiratory management protocols used in testing infants suffering from acute hyaline membrane disease (HMD) involves maintaining a delicate balance between meeting the infant's respiratory needs and avoiding the devastating long-term consequences of excessive airway pressures and oxygen. Systematic use of all the knowledge affecting ventilator therapy decisions will improve the quality of care, particularly during periods when the most experienced clinicians are not present. The objective of this project is to design and implement a computer-based system which will aid clinicians in providing appropriate respiratory support for infants suffering acute HMD. The system will build on our previously developed computer-based HMD course and management model. Knowledge of treatment protocols and the therapeutic and diagnostic states under which each protocol operates will be derived from our current clinical practices. This knowledge will be expressed as treatment and diagnostic rules and incorporated in a small knowledge-based system written in Turbo Prolog and implemented on a personal computer. Such an approach allows flexible use of qualitative physiological and clinical knowledge to an extent which is difficult to achieve with traditional programming methods. The resulting system can be modified and expanded to extend the domain of applicability to a broader population and improve the system's predictive accuracy. This system will use readily available data (e.g. blood gases, respirator settings, physical signs) in conjunction with diagnostic rules to maintain a list of attributes reflecting an individual baby's current physiological and treatment status. The system will search the treatment rules for all applicable therapy changes which could correct a given blood gas abnormality, and list the alternatives along with their side effects and the applicable rationale. It will be a valuable tool in teaching the principles of ventilator management. All details of the patient's state and each user interaction with the system will be stored and used in evaluating system performance. This system should enhance the consistency of respiratory support by consistently applying a large body of timely clinical knowledge to each therapeutic decision.
| Dawant, B M; Manders, E J; Lindstrom, D P (1994) Adaptive signal analysis and interpretation for real-time intelligent patient monitoring. Methods Inf Med 33:60-3 |
| Uckun, S (1994) Intelligent systems in patient monitoring and therapy management. A survey of research projects. Int J Clin Monit Comput 11:241-53 |
| Uckun, S; Dawant, B M; Lindstrom, D P (1993) Model-based diagnosis in intensive care monitoring: the YAQ approach. Artif Intell Med 5:31-48 |
