Every minute Google gets over 70,000 health-related searches. This amounts to 7 percent of Google’s daily queries. In most cases, people try to find out the possible cause of their symptoms and understand what to do next.
Sure thing, neither Google, nor popular online symptom checkers can replace visiting a doctor. However, when embedded into existing health systems and supervised by professionals, pre-diagnostic technologies may deliver a lot of advantages. And first off, they ease the burden on medical personnel — especially important in resource-limited settings and pandemics.
How do healthcare organizations quickly implement such tools in their daily practice? The answer is clear and short — via APIs (application programming interfaces). This article gives an overview of available APIs along with their primary use cases. But first, let’s examine the main parts and core functionality of symptom checkers that can be integrated into a hospital’s daily workflow.
Symptom checker API components and solutions you can build with them
Symptom checker APIs are not here to put real physicians out of work. Their mission is to keep patients better informed about the possible roots of their conditions and provide clinicians with decision support. Typically, symptom checker APIs include two major components.
A knowledge base. It contains data on conditions, diseases, and treatment procedures. The content is constantly reviewed and updated by medical professionals.
A diagnostic engine. Often powered by AI, it analyzes patient data inputs (like demographics, symptoms, and lab tests) or automatically extracts clinical features from electronic health records. The engine links patient information with pieces of content in the knowledge base and returns a list of likely conditions (preliminary diagnosis), care suggestions (triage), or both.
Using APIs, healthcare organizations and software providers can quickly build and integrate three major types of software solutions.
Triage tools. In medicine, the term triage means the quick sorting of patients according to the severity of their case. Triage tools facilitate this process and are typically used in nurse call centers and emergency departments
The World Health Organization (WHO) puts patients into three triage categories:
- emergency — for those who need immediate treatment,
- priority — for those who should be a priority for rapid assessment, and
However, triage systems can be more granular and have up to five levels of urgency.
Self-diagnostic apps or health chatbots. They help people discover possible causes of their symptoms and direct patients to the right specialist.
Pre-diagnostic decision support tools. This type of software is meant for professional use in hospitals, medical institutions, insurance companies, and so on. They can be integrated with telehealth platforms, EHR systems, hospital websites, and mobile apps.
Symptom checker API comparison
Usually, API vendors provide developers with detailed documentation and a testing environment. A diagnostic solution can be designed and integrated with the existing app or system in just a few days. Below you’ll find some popular options to try.
Infermedica API for diagnosis and decision support
Primary use cases: enriching existing healthcare solutions with diagnostic insights, building triage tools for hospital call centers, diagnostic chatbots and voice assistants, or advanced clinical decision support systems.
The Infermedica API offers an AI-powered diagnosis engine with natural language processing (NLP) capabilities and a broad knowledge base. It contains information on over 1,500 symptoms and 800 conditions, including pathologies, disorders, and diseases, related to COVID-19. The content is available in 17 languages.
The functionality can be used for patient triage, preliminary health assessment, and decision support. Upon receiving basic patient data — like symptoms, demographics, and lab test results, — smart algorithms analyze the information and return a list of possible conditions and relevant recommendations.
The API serves as a solid foundation for building a diagnostic chatbot as its NLP technology can capture symptoms in patient messages. You also can design a digital voice assistant: Infermedica supports three major voice platforms — Amazon Alexa, Microsoft Cortana, and Google Assistant.
The declared accuracy of the AI is 93 percent. To attain even better results, it is constantly extracting new data from millions of sources and reassessing the probabilities of different conditions using machine learning. At the same time, to comply with privacy protection laws, it never collects personal information .
Mayo Clinic API for symptom triage
Primary use cases: integration with the existing hospital or medical institution software to facilitate medical triage
The Mayo Clinic is renowned globally for its contribution to clinical practice, research, and science. Its knowledge base contains over 11,500 pieces of content, covering over 300 symptoms.
The symptom triage API runs AI algorithms that combine information from the knowledge base and real-time patient inputs to deliver care guidance using JSON or XML files. When integrated with a hospital app or website, the triage service may analyze questionnaires filled out by patients, suggest diagnoses, facilitate the decision-making process, and thus optimize emergency room visits.
ApiMedic API for patient guidance
Primary use cases: integration with consumer health apps and hospital websites to guide a patient’s journey
ApiMedic Symptom Checker prompts patients to enter their symptoms, generates a list of possible diseases, links them to medical information, and directs them to the correct doctor. The full functionality encompasses:
- calculation of disease likelihood,
- a long medical article per disease, and
- a red flag warning to indicate critical conditions.
The API integrates with websites and mobile apps with just a couple of lines of code. It can also be used for building a new application and supports several languages, including German, French, and Spanish. The developer portal provided by ApiMedic has a testing environment as well.
Isabel symptom checker API for generating a differential diagnosis
Primary use cases: adding diagnosis decision support functionality to EHR systems and hospital websites, building standalone symptom checker apps.
An AI-based differential diagnosis (DDx) generator by Isabel Healthcare covers over 10,000 diseases, from common to rare ones. It utilizes natural language processing to extract clinical symptoms from medical records. The tool allows users to input symptoms in common words — instead of selecting from a limited list of suggestions. When entering lab values, the NLP engine automatically translates them into everyday language — for example, “hemoglobin 4.2 mmol” will be converted to “low hemoglobin.”
Isabel returns a list of potential diagnoses relevant to a patient’s age, gender, and region they live. It flags the most critical diagnosis and also links each condition to the latest evidence-based clinical content about it. Information partners include DynaMed, 5MinuteConsult, and BMJ Best Practice.
Isabel API makes the functionality easy to embed into most EHR systems, as it comes pre-integrated with major players — such as Cerner, Epic, NextGen, T-System, SystemOne, and others. The suite of DDX features is available in XML and JSON formats.
EndlessMedical API for building symptom checker apps
Primary use cases: building websites, mobile apps or chatbots for hospitals, or integration with existing health systems
EndlessMedical API is created by doctors for doctors to connect a patient’s complaints with test results and findings of the professional examination.
After analyzing data, the ML algorithm returns a JSON response with a list of possible diagnoses, their likelihood rank, and flagging of life-threatening, emergency, or rare diseases. It also suggests additional questions to be asked, tests to be ordered, and specialist to be visited. From March 2020, the pre-diagnostic API has covered COVID-19 and can be used to triage patients with flu-like respiratory disorders, fever, and other coronavirus symptoms.
Health Navigator API for hospital triage and pre-diagnosis
Primary use cases: creating digital health assistants and chatbots, adding diagnostic functionality and NLP capabilities to existing EHR systems, telemedicine platforms, and healthcare applications.
Health Navigator API gives access to a suite of modules so that healthcare providers and medical software vendors can choose exactly what they need. The full set includes:
- a diagnosis engine to build a symptom checker for emergency-care, non-emergency, and self-care scenarios;
- an NLP engine to convert words patients use for describing their symptoms into medical codes and terminology that can be utilized by applications and care providers;
- a content database that contains a list of 7,200+ references and 25 000+ Internet resources linked to the 2800+ clinical concepts;
- a triage engine to be used in patient health bots and triage apps for emergency departments and call centers. The decision support tool runs a decision tree algorithm to make care recommendations based on such factors as patient complaints, age, gender, known medical problems, and more;
- a Coded Chief Complaints vocabulary to capture and store the reason for the visit in clinical documentation covering 470 patient complaints and supporting eight languages;
- a Clinical Documentation Support engine to ask patients the right questions during an encounter and create a documentation checklist; and
- After Care instructions to provide health information and care advice for telehealth patients.
Sonde Health API for respiratory symptom checking
Primary use cases: integration with iOS and Android apps to run medical triage in hospitals or population health checks and monitoring
Sonde Health pioneers a new type of diagnostic approach based on so-called “voice biomarkers” or subtle changes in the voice timbre. These biomarkers may be a sign of disease — from COVID-19 to heart ailments to depression.
The AI-fueled technology was created at the Lincoln Laboratory (MIT) and trained on one million voice samples obtained from hospitals and health providers over the last five years. It is capable of identifying over 4,000 vocal features with just a 6-second voice sample.
Now, the voice-based symptom checker is available for developers and organizations of all sizes via the HIPAA-compliant API. Due to its ability to detect respiratory conditions in just a few seconds, the solution can be used for daily screening of patients in hospitals or employees in the office. For example, one healthcare provider applies Sonde Health technology to check patients for signs of depression.
Symptom checker API benefits
Quick to set up, pre-diagnostic solutions yield benefits to medical staff and patients. Here is how a symptom checker can improve patient care and hospital workflow when implemented into daily practice.
- It helps physicians solve diagnostic dilemmas and encourages them to consider other possibilities.
- It reduces the likelihood of delayed or wrong diagnoses.
- It speeds up correct diagnosis which is a fundamental driver for clinical and financial performance.
- It makes patients more informed about their conditions and educates them.
- It facilitates the patient journey within the healthcare ecosystem.
- It reduces the number of unnecessary hospital visits.
- It optimizes the workload of emergency departments.
By no means, can symptom checkers be depended on for final decision making. It’s only a way to achieve a better diagnosis and as such these technologies work quite efficiently.
How to implement a pre-diagnostic tool into daily workflow
APIs simplify the technical side of software adoption, sometimes reducing the time needed to embed the new functionality into your current IT ecosystem to just a few hours. However, the success of innovations largely depends on their users — or, in this case, of healthcare workers and patients.
Below are key tips for technology adoption distilled from the experience of institutions that have made symptom checkers a part of their daily routine.
Introduce diagnostic tools to the staff early
Before the solution begins working, you should explain to your employees why the organization decided to adopt this software and what difference it is expected to make. Organize training, provide web tutorials, and make sure that clinicians are aware of where and who can access the diagnostic tool.
The good idea is to plan a launch event — once everybody is on board. This will deliver a clear message on when to start using the tool.
Ensure ease of access
Multiple and easy to find points of access play a key role in promoting new technologies. Integration with an EHR system makes symptom checkers available directly when and where physicians are making diagnosis decisions. But availability from a clinical portal, hospital website, or app will speed up the tool adoption as an intrinsic part of daily workflow.
The success of adoption can be measured by quantitative changes in readmissions and test orders or by patient satisfaction scores. You also should collect feedback from clinicians who work with the tool in the first place — including physicians, nurses in call centers, and triage officers. This will help you finetune the ways your personnel use the technology so that everybody gets most of the system.
Originally published at AltexSoft blog “Symptom Checker APIs: How They Improve Medical Triage and Diagnosis”