Digitizing and extracting information from handwritten documents is a particularly tricky challenge within the larger challenge of processing differently formatted documents.
Whether a scanned handwritten document or a photo of a handwritten note, OCR solutions come up short because there is such a wide range of handwriting that programmers can’t provide enough examples of how any one character should look. Even AI hasn’t been able to master handwriting recognition and extraction to perfection–if a person has difficulty interpreting someone’s handwriting, how can we expect a computer not to?
Layer over that an application with the potential to save or end lives–medical prescriptions–and you have a high-stakes, high-risk situation. Not to mention doctors have notoriously sloppy handwriting, which is then left to the pharmacist to correctly decipher. When there’s a translation error in this process, who, then, is responsible?
In the US in the 1960s, patients began to litigate medication errors in consumer courts. Once hospitals were found liable and were forced to pay compensation, they invested in pharmacy infrastructure, hiring more qualified pharmacists with advanced degrees–an expensive upgrade, but considerably less than what they were paying in settlements.
Today, almost all countries have some legislation in place mandating a shift from handwritten to electronic prescribing. Digitizing patient records and prescriptions helps solve the medication-related problems that illegible handwriting causes, but of course there are always exceptions. In the United States, for example, the federal government has required e-prescribing for controlled substances for patients using Medicare, but further legislation is left to the discretion of each individual state.
Prescription processing boosted with the power of collective intelligence
Crowdsourcing can offer a solution to the pains of processing prescriptions, whether handwritten or electronic. The practice of microtasking—breaking a large task into multiple small tasks that can be worked on simultaneously—is essential to both.
ScaleHub collective intelligence solutions snippet an image or document into many tiny pieces before sending it to the crowd for processing. We send every snippet to two different crowd contributors simultaneously; in the case that the results don’t match, we’ll send the same snippet to a third, more experienced contributor for a tie-breaker. This is the secret sauce behind our guaranteed 99% accuracy rate—even for handwritten text.
Prescription processing service providers often handle both order processing and the financing around those orders. It goes without saying that these providers are held to very demanding SLAs around accuracy, speed and data privacy. Volume volatility under these conditions threatens their ability to meet their SLAs, as we’ve seen over the past few years due to the pandemic.
With collective intelligence based on crowdsourcing, prescription processors can more easily manage volume spikes or dips while meeting SLAs—without hiring additional staff. If a mistake were to occur somewhere between a doctor prescribing a medication and a patient consuming it, a crowdsourcing collective intelligence portal offers the added benefit of an electronic processing footprint. The source of the error—regardless of where in the treatment process it occurred—can easily be traced.