How Artificial Intelligence Can Help Increase Patient Safety

Introduction

The basis of the clinical practice has been very well established in the last two centuries, but the Artificial Intelligence (AI) could modify substantially this situation. We would like to challenge the actual diagnostic paradigm with a particular clinical case, enhancing the contribution of this tool to patient safety.

Clinical Summary

We show here a simple chest x-ray from a 62-year-old woman, with a previous exposure to asbestos. He started with a progressive shortness of breath and a left thoracic pain, without fever or sputum.

You can see in the image on the left side, an opacity, suggesting a pleural origin. Once we rule out pneumonia (no fever or sputum) or an atelectasis (there is no tracheal deviation to the left), then the most probable diagnosis is a Malignant Pleural Mesothelioma.

Differential Diagnosis

If we asked for an AI tool about the diagnosis of this patient, the answer with almost total accuracy will be a pleural Mesothelioma, but we will be surprised when you see  the amount of useful information widening the list of clinical entities, some of them very uncommon in the daily clinical practice. We will show the list of possible diagnoses, supplied by ChatGPT.

Question to ChatGP:  If you thought as a medical oncologist, What could be the possible diagnosis for a man with a pleural oppacity, and previous exposure to asbestos?

. Malignant Pleural Mesothelioma

.Asbestos-associated lung cancer (exposure to asbestos increases the risk of developing adenocarcinoma and squamous cell carcinoma)

. Benign asbestos pleural disease: pleural plaques and/or benign pleural effusion

.Benign pleural tumors such as pleural lipoma and solitary fibrous tumor of the pleura.

.Pleural Metastases

Comment

One of the advantages of consulting an Artificial Intelligence tool is the speed to obtain the results, also using a mobile phone. We can also use a “natural” language, without having to search for indexes in databases. But, in my opinion, the main advantage is the amount of data obtained and the list of clinical possibilities, some of which are uncommon.

In this case, the tool suggests some possibility such as “fibrous tumor of the pleura”, an entity that a general clinician would hardly think of, when making a differential diagnosis. At the end, what we are doing is to be sure that we have a proper and complete differential diagnosis, and with this, we are increasing patient safety.

Final Diagnosis

Malignant Pleural Mesothelioma

Author: Lorenzo Alonso Carrión

FORO  OSLER