1. Introduction

Advancements in AI for Health

In recent years, artificial intelligence has revolutionized various sectors, and healthcare is no exception. AI-driven tools have demonstrated remarkable capabilities in detecting diseases, predicting health outcomes, and personalizing treatment plans. For instance, deep learning algorithms have achieved dermatologist-level classification of skin cancer from images[^1], and AI models have been used to predict cardiovascular diseases using wearable device data[^2]. There are a multitude of other specialized AI classification algorithms covering tens of different diseases based on a variety of data inputs (skin image, text notes, wearable data, cough sounds, genetics, Xray images, genetics, lab reports, etc)[^3].

Additionally, Large Language Models (LLM) have enabled the development of conversational agents that can assess symptoms and provide preliminary health advice. A recent comparison study found that ChatGPT performed better than doctors in specific diagnostic tests [^4]. AI has also shown promise in mental health, with models capable of detecting signs of depression and anxiety through voice and text analysis[^5].

Our commitment is to provide these services with the highest ethical standards. We believe that individuals should maintain control over their health data, benefiting from its use and only with their informed consent. We hold that people always own their data and should, therefore, have a share in any value generated when they give others access to it.

We pledge to use our platform to serve people in a way that enhances their lives and promotes better health outcomes globally.

Barriers to Accessibility

Despite these advancements, access to AI health tools remains limited for the general public, especially in emerging countries and rural areas. The barriers include:

• Fragmentation: Many AI models are confined to academic research or niche proprietary platforms (one app for one disease), making them inaccessible to those who need them most and hard to advertise.

• Regulatory Hurdles: Lengthy and costly approval processes and varying international regulations delay the deployment of AI health tools. The lack of clarity on which AI algorithms can be considered a medical device exacerbates this issue.

• Cost and Infrastructure: High costs can limit the use of these technologies in low-resource settings. Many people around the world lack the ability to pay for these services using credit cards. Additionally, some of the existing tools are only made available in limited geographical locations (For example: SkinVision).

The Potential Impact

Making AI health tools accessible can have a transformative impact:

• Timely Interventions: Early detection and intervention can prevent diseases from progressing, reducing morbidity and mortality rates. AI-powered smartphone tools have demonstrated significant potential, particularly in emerging countries and rural areas with limited access to specialized care. For instance, by screening thousands of infants in underserved regions, preventing irreversible blindness by enabling timely diagnosis and treatment [^6].

• Empowering Individuals: Easy access to more health information empowers people to make informed decisions about their health. Having access to more information can also lessen the sense of helplessness and anxiety, especially while waiting to be seen by healthcare specialists.

• Reducing Healthcare Burden: AI tools can alleviate the strain on healthcare systems by providing preliminary assessments and triaging patients.

Carthalis aims to bridge the gap by providing a unified, accessible platform that hosts a multitude of AI health tools, ensuring that everyone, regardless of location or socioeconomic status, can benefit from the latest advancements in healthcare technology.