For patients who have a rare disease, it can take a long time to get an accurate diagnosis. Some conditions have doctors working months or even years to try to make sense of symptoms and test results. Patients cannot afford to forgo treatment while doctors try to discern clues about their conditions. For many, there’s no time to waste, yet it can take five months to a year or more to reach a diagnosis.
Cloud Therapy was founded to help doctors distill diagnostic insights from enormous amounts of medical literature and research. The solution uses natural language processing (NLP), probabilistic reasoning and deep learning to analyze terabytes of unstructured data to assist doctors recognize rare conditions quickly and accurately.
If the solution can help medical professionals access information faster, it can mean the difference between life and death for someone with a rare disease.
The predecessor to Cloud Therapy was a simple iPad app that contained information for patients to review before the doctor entered the consult room.
Throughout the day, a doctor may lose five minutes with one patient and six minutes with another while repeatedly answering the same questions. Going over the allotted time could eat up three hours, and the doctor would have to put in extra time to finish daily work. The app helped cut three down to about an hour and a half.
Around that time, IBM Watson debuted on Jeopardy!, and soon after, the technology became available for developers. This is when work began on the Cloud Therapy solution.
When the full range of Watson application programming interfaces (APIs) became available on the IBM Bluemix platform, Cloud Therapy took off.
Developing with Bluemix gives Cloud Therapy a competitive advantage because it can incorporate different services and data sources without spending time to develop custom integration logic. The company began experimenting with big data analytics and cognitive computing services. The team realized they had a noteworthy idea when the Global Mobile Innovators Tournament, enabled by IBM, chose Cloud Therapy as a finalist. The company’s prototype idea was later shared at the “4 Years From Now” (4YFN) conference in Barcelona, Spain.
Since then, the product has evolved, incorporating IBM Watson Conversation, Dialog, Natural Language Classifier, Retrieve and Rank, Speech to Text and Text to Speech services along with the IBM Cloudant noSQL database, IBM Mobile Push Notification and IBM Mobile Quality Assurance services.
Cloud Therapy relies on Cloudant because the company wants people to have the best possible experience with no lag. Delivering the solution in seconds is what makes Cloud Therapy what it is.
With IBM Bluemix Mobile Services, the Cloud Therapy solution supports mobile app capabilities, including the ability for users to offer immediate feedback within the app. It also uses Google Translate to accommodate English and Spanish language content.
Many might wonder if they could use a regular search engine to understand the relationships of the question the way Watson does. With a search engine, people must cross your fingers that whatever results they get will be meaningful. The Cloud Therapy solution uses Watson’s brain to understand relationships and actually learn.
For each customer implementation, Cloud Therapy cleans, curates and uploads data, including clinical trials, drug research, previous case information, and genotype and phenotype patterns. The data can be both structured and unstructured. Initially, the company creates question-and-answer pairs to train the algorithms. The training process involves providing feedback to the system as it matches appropriate answers to questions. This requires input from both subject matter experts and data scientists.
After learning from the data set, the cognitive engine uses NLP to understand queries and then applies textual analysis and probabilistic reasoning to return appropriate content, ranked by confidence level. The platform can process 1.3 terabytes of data per second.
The solution also learns from user feedback. Doctors can converse with the system in natural language, and it helps provide clues to a patient’s condition.
For several pilot implementations, each Cloud Therapy customer brings its own data and desired use case. For one project, a customer will provide 13 years of data on 9,000 rare disease cases that doctors will want to query.
Cloud Therapy tells all its customers that this is experimental technology and they should understand that, especially because expectations in the cognitive era are enormously high. The company tries to help everyone recognize that the technology is brand new, and so far, so good.
Cloud Therapy sees the value, and its partners see the value. Understanding patient symptoms for a faster, more accurate diagnosis can be a matter of life and death.
You must be logged in to post a comment.