Can AI Help Sudan’s Collapsing Healthcare System?  

Sudan’s healthcare system, already strained before the civil war, is now on the verge of collapse, with only 30% of facilities operational and widespread shortages of medical staff and supplies. AI offers potential solutions by enhancing diagnostics, automating administrative tasks, and supporting remote consultations, but challenges such as infrastructure limitations, algorithmic bias, and data reliability must be addressed for effective implementation.

Sudan’s Healthcare Crisis: A Call for Immediate Solutions 

Sudan has been embroiled in civil war for nearly two years. A November 2024 report by the London School of Hygiene and Tropical Medicine’s Sudan Research Group revealed that the conflict has claimed over 60,000 lives in Khartoum State alone, with more than half the population facing starvation. 

The ongoing war has devastated Sudan’s healthcare system. The World Health Organization (WHO) warns a collapse, citing insufficient medical supplies, loss of healthcare workers, and destroyed facilities.  

Only 30% of health facilities remain operational, with medical supplies meeting just a quarter of the demand. This dire situation has fuelled the spread of diseases, and people are often unable to access healthcare due to the dangers of venturing out during the conflict. Over 1.3 million cases of malaria, 11,000 cases of cholera, and 4,600 cases of measles have been reported.  

To address these challenges, Sudan urgently needs innovative healthcare solutions that can alleviate the strain on traditional medical system. 

AI in War-Torn Healthcare Systems: Opportunities and Obstacles 

Al-Moghirah Al-Amin Gad Al-Sayed, Federal Ministry of Health in Sudan, emphasized that health authorities are increasingly turning to AI to tackle shortages of medical personnel and resources. AI’s capability to analyze medical data, such as X-rays, and generate diagnoses without the need for human interference could prove invaluable in war-torn regions with limited access to physicians.  

AI has already proven useful in enhancing medical assessments across various healthcare fields, including pathology disease and pulmonary embolism detection.  

“Many studies have proven that AI shows great accuracy in medical diagnosis as a result of the data on which it has been trained, as well as the summary of human experiences and experiences that have fed it” 

Al-Moghirah Al-Amin Gad Al-Sayed, Federal Ministry of Health in Sudan. 

Globally, AI has demonstrated its potential in healthcare and is increasingly being explored in the context of war-torn environments. For example, following the onset of the Syrian conflict in 2011, the Syrian American Medical Society utilized digital tools such as WhatsApp and video technology to support remote ward rounds in ICUs with specialists based abroad. Although effective, this approach could have been significantly enhanced with AI integration. 

In 2019, a U.S.-based healthcare organization implemented an early detection system for sepsis, called Sepsis Prediction and Optimization of Therapy across 173 hospitals, reducing severe sepsis mortality by nearly 10%. Such technologies can also be implemented in war-torn countries, where timely detection and intervention could save countless lives in the absence of sufficient medical personnel. 

Medical imaging, a critical component of modern diagnostics, can benefit tremendously from AI. Tools such as Aidoc, Viz.ai, and Zebra Medical Vision have demonstrated capabilities in detecting conditions like stroke, pulmonary embolism, and fractures with remarkable accuracy. However, challenges remain. 

A study suggests that Aidoc, have been found to miss diagnoses or produce false positives. Such errors could be catastrophic in Sudan and other war-torn countries, where patients may lack access to second opinions.  

AI Chatbots and Combating Antimicrobial Resistance  

Antimicrobial resistance is a major healthcare challenge in Sudan, driven by medication misuse stemming from insufficient medical guidance. AI chatbots, like ChatGPT, could provide remote medical advice in areas lacking healthcare workers. By guiding patients on appropriate antibiotic use, these tools have the potential to curb antimicrobial resistance in conflict zones. Moreover, digital clinicians such as those provided by Hippocratic AI may also be beneficial in these situations.  

However, concerns about the accuracy of AI-generated medical advice persist. Last year, an Austrian donation-funded NGO, called None of Your Business, worked to enforce data protection laws and issued a formal complaint after OpenAI admitted its inability to amend inaccurate information provided by the chatbot. This incident underscores significant concerns about the reliability of AI-generated medical advice, which could exacerbate Sudan’s already collapsing healthcare. 

AI in Administrative Tasks: Reducing Strain on Medical Staff 

AI tools have shown promise in streamlining administrative tasks, shortening the time needed for each patient interaction.  

For example, medical scribing tools such as Suki, AmazonOne, and Whisper can prove invaluable in tackling the clinician shortages in Sudan. Moreover, AI-generated medical hand-off notes have proven useful in emergency medicine. When applied in Sudan, it could reduce miscommunications between clinicians and ensure that no patient was missed or misdiagnosed.  

Nevertheless, these tools are not infallible. Whisper has been found to inaccurately record medical information, which given Sudan’s limited resources, such inaccuracies could further strain the healthcare system. 

Use of AI for Cancer During Humanitarian Crisis 

At the 2024 World Cancer Congress (WCC), discussions emphasized the importance of addressing the increasing challenges of cancer screening, diagnoses, and treatment in places such as Sudan, Gaza, and Haiti.  

For instance, a study involving more than 24,000 women, evaluated an AI tool, called, Automated Visual Evaluation (AVE) for cervical pre-cancer screening. It was used in five countries, where oncology systems are not advanced —Malawi, Rwanda, Senegal, Zambia, and Zimbabwe. The AI tool showed a significantly higher sensitivity of 60.1%, compared to traditional methods.  

“The study concludes that AVE could enhance cervical pre-cancer detection in resource-limited settings, supporting broader cervical cancer elimination efforts despite some trade-offs in specificity,” 

Prof Jeff Dunn, President of the Union for International Cancer Control (UICC). 

AI’s potential extends beyond diagnosis and administration. In humanitarian contexts, AI can improve the early detection of health threats and improve logistical planning. The UNHCR’s Project Jetson, for example, used predictive models to forecast forced displacement in Somalia, demonstrating AI’s ability to anticipate outbreaks and manage healthcare resources effectively. 

Can AI Replace Human Clinicians? 

Although AI holds immense potential, it cannot fully replace the critical need for additional healthcare workers, medical supplies, and infrastructure in Sudan. As Al-Sayed noted, while AI might alleviate some burdens, the country still requires significant investment in human resources to address its healthcare crisis effectively. 

The reliability of AI tools depends heavily on the quality and selection of their training data. A study evaluating a Korean AI tool designed to detect malaria found that it required expert intervention when applied in different regions. This highlights a key limitation: AI trained on data from other regions may not perform optimally in Sudan. However, in a country with limited resources, collecting high-quality, representative data poses a significant challenge. 

Algorithmic biases present another critical issue. These biases often arise from training datasets that overrepresent certain groups, while underrepresenting others, perpetuating historical inequities. For example, a study found that AI tools struggle to accurately diagnose skin conditions across diverse skin tones due to insufficient representation in training datasets. Such disparities could be especially problematic in Sudan, where the population is predominantly Black African. 

The applicability of AI tools in Sudan’s healthcare system is not the only challenge, implementing and using these AI tools may also prove difficult, due to factors including: 

• Infrastructure limitations: unreliable internet, power interruptions, and lack of equipment. 
• Data management: fragmented health information systems and poor patient data management for the AI tools to use. 
• Workforce readiness: insufficient training for healthcare workers to interpret AI outputs. 
• Ethical concerns: risks related to data privacy, algorithmic bias, and potential politicization of healthcare data. 
• Regulatory hurdles: weak legislative frameworks and inconsistent guidelines. 

These factors underscore that even the most advanced AI tools may falter when faced with the logistical and systemic difficulties within collapsing healthcare systems, like that in Sudan. However, despite this, AI still serves as one of the best innovations that can be used to alleviate the monumental loss of the health workforce.