AI in Oncology

AI has made significant inroads into various medical specialties, with oncology being the latest beneficiary of this powerful technology. To this end, in 2020 there were more than 60 Food and Drug Administration (FDA) approved medical device and algorithm solutions across a variety of therapeutic areas.

Read on to learn more on how AI in oncology is positively transforming cancer diagnosis and treatment and how AI bias can be resolved.

Cancer Impact

A cancer diagnosis brings unprecedented economic and social change to affected individuals and families. There’s understandable pain, fear, and confusion. As a healthcare community, it’s important for us to find solutions that support the best possible patient outcomes. 

Cancer treatment costs are high, even when individuals are covered by insurance, there’s also a potential loss of income during treatment. Some insurance policies may not cover all related cancer costs, and patients can end up with large out-of-pocket expenses. 

Finances aside, cancer breaks the social fabric. Relationships are tested in the ensuing confusion, stress, and sadness. A cancer diagnosis may lead to depression and feelings of defeat, along with many other stressful emotions.

Numerous studies show when cancer is diagnosed in the early stages, patients have better outcomes. It’s therefore critical for the healthcare community to do all they can to support the early diagnosis of cancer. One way to support this effort is through a second, safety net read of all radiology imaging – ensuring no potential or early-stage cancers are missed. 

Current Cancer Screening Processes

Presently, cancer screening involves physical exams, genetic tests, laboratory tests and imaging procedures. A review of the patient’s medical history is also needed for a conclusive diagnosis. 

While these tests are crucial in identifying cancer, another problem persists: misdiagnosis. Due to factors like inexperience, fatigue, limited peer review or bias, some doctors may form a wrong diagnosis, potentially prompting further medical procedures to the patient’s detriment.

Other additional factors known to inhibit effective cancer screening include:

  • Technical failure
  • Communication failure between medical professionals
  • Workflow and staffing pattern
AI in oncology; lung cancer radiology CT scan

Misdiagnosis and other medical errors may be easily waved off as minor issues, especially because these misreads aren’t intentional – they’re errors of omission that were not committed on purpose. Still, it’s also one of the leading causes of death in the US. The rate of misdiagnosis is estimated to be between 10% and 15%, but only less than 10% is reported according to PubMed.

Even with a low incidence rate, the high number of people who get injured or die each year due to misdiagnosis suggests that corrective action should be taken fast. This is where smart AI in oncology solutions can support physicians in their daily practice.

Aside from the possibility of injury or death, patients may be forced to pay for inaccurately prescribed medicines or procedures. 

Overview of AI in Oncology

The uptake of AI in oncology hasn’t been so much despite the technology showing much promise. We believe this is due in part to an overwhelming amount of data being reported by these tools that is causing physicians and their staff to be bogged down the need to decipher and coordinate the data together. That’s what we’re trying to collect and build — better big data so all doctors can have all details ready at their fingertips to provide the best holistic care they can

Among other things, oncology AI can improve:

  • Communication among clinicians
  • Cancer diagnosis and treatment
  • Speed of workflow
  • Cancer research

Even though AI has proved it can tackle bias, it can still occur if your AI solution is not validated on a hospital’s own patient population. AI in healthcare overall has the opportunity to boldly change how health systems ensure they are delivering the right care for their patients, from population screening through diagnosis and treatment. There is the opportunity for AI solution providers to establish deeper partnerships with health systems where, rather than selling AI applications one by one to different clinical leaders. Ideally, physicians leverage a AI tool at the system level that can fundamentally revisit how patients flow through their system and drive deeper systemic change together.

Top Ways to Improve Screening & Diagnosis With AI

AI can revolutionize cancer screening and diagnosis in many different ways, such as:

Errors of omission and bias are found across healthcare, and in radiology scans, this could include a completely missed diagnosis or a delay in diagnosis and treatment due to discrepancies and undocumented findings. AI used in oncology scans supports providers and enhances patient care by achieving new and superior levels of patient analytics for diagnosing.

For example, lung cancer is the second most commonly diagnosed cancer and has risen to one of the leading causes of death in women. Lung nodules are often found incidentally on chest and abdominal CT scans, where the physician has the opportunity for early lung cancer detection. However, the existing peer review process, where lung nodules are identified early via a second scan review, is limited to a random screening of only 3% to 5% of all scans completed at a facility. One large health system recently implemented an AI platform to deploy a natural language processing algorithm to augment the work of the radiologists. Over the course of eight months:

  • the platform scanned 30,000 images
  • 1,137 were flagged for missed findings
  • 118 new cases were found to have clinically significant lung nodules, directly impacting the course of patient care

Based on this study and others, the use of AI to identify undocumented lung nodules supports early lung cancer detection and improved diagnosis.

Imaging classification is a crucial part of forming any diagnosis. However, with years of experience and skill, imaging classification can still take a long time when done by doctors – because it’s subjective. 

The use of AI in oncology provides medical professionals access to data patterns that help them better understand patient data that in turn directly impacts care. With the adoption of AI, providers can make more informed decisions, arrive at solutions faster, and assess medical images and records with more accuracy. It also helps prevent fatigue which may also contribute to misdiagnosis.

A women’s health example exists in breast cancer detection. By using artificial intelligence in oncology, medical professionals can quickly and accurately sort through breast MRIs in patients with dense breast tissue to eliminate those without cancer because the technology can do a lot of the upfront work to support the doctor.

Misdiagnosis of breast cancer can often occur because mammograms are less sensitive in women with extremely dense breast tissue than fatty breast tissue. Introducing AI to read breast scans helps bridge the reviewing process between the two types of tissue and streamline the diagnosis process.

Health equity may be defined as systematic differences in the health status or the distribution of health resources between population groups. Social factors such as education, employment status, income level, gender, and ethnicity all play a role in health outcomes

Clinicians may be unintentionally biased in their diagnosis and prescriptions, sometimes willingly and often without knowing at all. A well-developed cancer AI solution that is validated on your patient population can form unbiased cancer diagnoses and treatment prescriptions since it relies on a wide pool of datasets that include people from different demographics.

When implementing AI, it is important to find solutions that have been trained and tested on a diverse patient population, and an AI partner that can grow with you as your needs change. AI offers the ability to better understand inequities and ultimately help solve for them. 

The Future of AI in Cancer Diagnosis & Treatment

Behind all the incomprehensible technology, the main goal of AI is to improve cancer diagnosis and treatment for a better patient experience and a pleasant work environment for the amazing clinical and administrative personnel.

Considering the dynamic nature of cancer treatment, the right AI vendor should offer:

  • Regular algorithm monitoring and necessary updates
  • AI algorithm validation on relevant patient population and equipment used for imaging
  • Top-notch data security
  • Capability to manage many use case requests at a reasonable cost
  • User-friendly interface, regular training and system troubleshooting support

What Clinicians Should Know

AI is a technology that can radically improve the quality of cancer diagnosis and how clinicians work. Therefore, it should not be viewed as something that’ll replace clinicians; rather, it’ll help them improve their patient’s experience and work experience.

How To Build an AI Oncology Strategy

Here’s the best way to build AI oncology strategy for your hospital or practice:

You should be able to know what AI is, its capabilities and limitations, and whether or not it can help your team with their work and improve the patient experience.

It is important to learn what they could use to help with the most and identify and address any concerns relevant stakeholders may have with AI acquisition and implementation in clinical operations.

Not all challenges can be solved using AI, and you should identify those that can be resolved using AI.

Identify the costs of AI implementation into your workflow, and check whether the benefits outweigh the costs. This might help justify your AI implementation strategy.

You can determine your desired return on investment from the project at this stage.

  1. Identify AI vendors that can offer the best solutions to your challenges and provide consistent support to your team.
  2. Develop a budget
  3. Acquire and implement AI into your workflow
  4. Track your work metrics for any changes since AI integration into your workflow.

AI bias can always reoccur at any stage of an AI lifecycle. It’s, therefore, crucial to work with credible vendors who will always guarantee AI updates that tackle AI bias and extensive AI validation. We welcome you to contact us for the best AI services. 

Key Takeaways

AI is one of the most talked-about technologies in healthcare. There are numerous areas of adoption, from powering surgical robots to enabling early cancer detection. In radiology, AI can increase the value practitioners provide their patients, thus improving healthcare outcomes. It can also streamline workflows and lessen the radiologists’ administrative burden.

Contact Ferrum Health to learn more.


Request More Information

Fill in the details below to request more information or a demo of our AI platform. Our team is standing by and ready to help!