By Author
  By Title
  By Keywords

June 2023, Volume 73, Issue 6

Student's Corner

Are we ready to adapt artificial intelligence for early sepsis detection?

Muhammad Moiz Nasir  ( First Year MBBS Student, Dow University of Health Sciences, Karachi, Pakistan. )
Moeez Tariq  ( Fourth Year MBBS Student, Dow University of Health Sciences, Karachi, Pakistan. )

DOI: 10.47391/JPMA.7978

 

Madam, Sepsis is a condition which entails life threatening organ dysfunction due to poorly regulated host response to infection1. It is a common cause of both morbidity and mortality1. Males and females both are at risk of death from sepsis, with males being at an increased risk2. The cause is variable but respiratory tract and urinary tract infections commonly lead to sepsis in Pakistan2. The diagnosis of sepsis has always been a challenge to physicians, internists, and critical care specialists. According to more recent guidelines, sepsis can be diagnosed by a combination of suspected infection and two or more points in the Sequential Organ Failure Assessment (SOFA) score1. SOFA assigns 1 to 4 points to each system working in the body, such that 24 is a composite and optimum score, higher scores being associated with increased mortality3. SOFA score is complex and time taking prompted the development of a quick SOFA (or qSOFA) for the earlier diagnosis of sepsis which includes respiratory rate of 22 breaths/minute or more, Glasgow coma scale score of less than 15, and systolic blood pressure of 100 mm of Hg or less. Patients having met any two of the qSOFA criteria along with suspected infection can be managed for sepsis1.

Recent advances in biomedical computation potentiate early sepsis detection through Artificial Intelligence (AI) by granting computers and machines power to simulate the problem-solving and decision-making capabilities of the human mind by utilizing computer science and robust datasets. These datasets are mainly available as structured and unstructured clinical data in the Electronic Medical Record (EMR)4. Structured data (such as heart rate, systolic blood pressure, diastolic blood pressure) and unstructured data (such as free-form clinical notes or radiological images) registered by nurse, lab reports and prescribed medications are inputted as data streams are used to create Machine Learning Algorithms such as “Sepsis early risk assessment (SERA)” algorithm developed by K Huat Goh et al.4 and “Early Warning System (EWS 2.0)” derived and validated by H M. Giannini et al.5 which allows both early detection and diagnosis of Sepsis. SERA algorithm potentially flags patients with sepsis that physicians may have overlooked in stressful hospital environments, therefore effectively flagging 21–32% more patients than clinicians 4-48 hours before sepsis onset with fewer chances of false positive results, as seen in figure4.

 

 

Sepsis affects 30 million people yearly, causing around six million deaths, with mortality rates reported as high as 30-40% in South East Asia6. Many of our population fluxes into public sector health care manage their illnesses. Provided the huge burden on the health sector, sepsis and its consequences remain a frequent cause of death, especially in medicine wards and intensive care units. Surviving Sepsis Campaign (SSC) guidelines provide a standardized approach to sepsis management7; however, diagnosis and early detection of sepsis is often an undervalued life-saving dimension where potential research investment such as in AI Detection Systems could generate clinically significant outcomes i.e. it will help our health care settings not only in early diagnosis but also aid in the prompt recognition of critical downfalls during the management of sepsis to prevent significant morbidity and mortality.

 

Submission completion date: 26-09-2022

 

Acceptance date: 18-01-2023

 

Disclaimer: None to declare.

 

Conflict of Interest: None to declare.

 

Funding Sources: None to declare.

 

References

 

1.      Evans T. Diagnosis and management of sepsis. Clin Med (Lond) 2018; 18: 146-9.

2.      Nasir N, Jamil B, Siddiqui S, Talat N, Khan FA, Hussain R. Mortality in Sepsis and its relationship with Gender. Pak J Med Sci 2015; 31: 1201-6.

3.      Walker BR, Colledge NR. Davidson's principles and practice of medicine e-book: Elsevier Health Sciences; 2013.

4.      Goh KH, Wang L, Yeow AYK, Poh H, Li K, Yeow JJL, et al. Artificial intelligence in sepsis early prediction and diagnosis using unstructured data in healthcare. Nat Commun 2021; 12: 711.

5.      Giannini HM, Ginestra JC, Chivers C, Draugelis M, Hanish A, Schweickert WD, et al. A Machine Learning Algorithm to Predict Severe Sepsis and Septic Shock: Development, Implementation, and Impact on Clinical Practice. Crit Care Med 2019; 47: 1485-92.

6.      Arshad A, Aziz A, Ayaz A, Salahuddin SM, Jamil B. Knowledge and attitude towards identification and management of sepsis among resident physicians in a tertiary care teaching hospital in Pakistan. J Pak Med Assoc 2021; 71: 1000-1.

7.      Evans L, Rhodes A, Alhazzani W, Antonelli M, Coopersmith CM, French C, et al. Surviving sepsis campaign: international guidelines for management of sepsis and septic shock 2021. Intensive Care Med 2021; 47: 1181-247.

Journal of the Pakistan Medical Association has agreed to receive and publish manuscripts in accordance with the principles of the following committees: