Threshold for Defining Fever Varies with Age Especially in Children and can be Further Improved Using Artificial Intelligence Techniques: A Multi-site Diagnostic Accuracy Study | Chapter 15 | Emerging Trends in Disease and Health Research Vol. 7

 Fever is a key prognostic indicator of disease. It is a symptom of viral infections, but it can also be a symptom of bacterial infections that can be treated with medicines. In general, the risk of infection rises with the severity of the fever, although infection is the root cause of fever and the most dangerous aspect of disease. As a result, while determining a fever threshold, the optimal fever threshold should be based on a comparison of the temperature measurement to "real disease or infection as determined by diagnostic tests and a thorough patient examination by a physician or health care professional."

Fever is defined by the American Academy of Pediatrics and the European Centre for Pediatric and Adolescent Medicine as a temperature of more than 38.0°C in people of all ages. While the AAP and ECPA recommend a stable temperature of 100.4°F (38.0°C) as a guideline, it delivers a poor prognosis for infections and illness. Herzog et al [1] adopted a different approach, doing an exhaustive examination of the literature to investigate the various cutoff points and determine the lower limit of "fever" and "severe fever" based on the age of the patient.

Design: A multi-site diagnostic accuracy research was undertaken on a total of 894 individuals, 373 of whom were sick, to compare a 'age-based' threshold model with a 'fixed' threshold over 38.0°C.

Methods: A clinical categorization ("healthy" or "sick") completed by a doctor through a comprehensive examination was compared to the 'age-based' and 'fixed' threshold fever determinations.

Results: In all ages, the sensitivity and accuracy of age-based thresholds were found to be superior to set thresholds. Using an ensemble decision tree based Artificial Intelligence system with age and numerous other parameters, the sensitivity and accuracy were shown to improve even more.

Conclusion: Our findings showed that the empirical model proposed by Herzog et al [1] for age-based fever thresholds showed a closer agreement (in terms of sensitivity and accuracy) between fever due to elevated temperatures and illness as identified by a clinical impression from a Health Care Professional. This agreement was also improved by the AI model utilising a decision tree ensemble approach.

Clinical Importance: The framework given by this study will help parents and caregivers make better decisions about whether or not to seek medical help [2]. It will also enable for the treatment of fevers at home rather than having to go to the doctor. As a result, parents' medical bills will be lower, and medical resources will be used more efficiently. Other authors [3], for example, have showed a large cost savings as a result of precise prognosis of newborns with illness or infection.

Author(S) Details

Rajesh S. Kasbekar
Helen of Troy, Inc., Regulatory and Clinical Affairs Department, 400 Donald Lynch Boulevard, Suite 300, Marlborough, MA 01752, USA.

View Book:- https://stm.bookpi.org/ETDHR-V7/article/view/6602