Since the start of 2013, artificial intelligence-focused startups have received over $66 billion in funding, and that number is already likely higher by the time you’re reading this. BenchSci has a great list of over 200 life sciences-focused companies using AI (including us!). You’ve probably noticed the “.ai” at the end of our URL, indicating that we, like many other companies, are incorporating artificial intelligence in our platform.
But what does “incorporating artificial intelligence” actually mean? How does LabVoice use AI to propel its virtual assistant’s platform forward? Oftentimes, companies have vague answers when it comes to their use of AI or ML, so we’ve decided to list some concrete examples of how LabVoice uses AI to create a more interactive, digital lab experience.
LabVoice is Process-Aware
Experiments, equipment issues, and other lab processes don’t always proceed as planned; lab work is filled with deviations. Sometimes those deviations are minor bumps in the road, and other times they’re roadblocks, requiring great time, effort, and resources (read: people) to overcome them.
LabVoice has built a process-aware, adaptive system that can account for many deviations ahead of time, and provide the user with a set of instructions on how to move through those roadblocks efficiently and not necessarily requiring anyone else’s time.
Let’s use the below example: the user instructs LabVoice to collect the pH levels, and LabVoice checks against the calibration records to see if the meter has already been calibrated.
On the left-hand side, LabVoice confirms that a specific meter is ready to be used, and proceeds to follow the user’s instructions.
On the right, LabVoice catches that the meter has not been calibrated, and asks the user if they’d like to calibrate the instrument. Once the balance has been calibrated, that user picks up right where they left off. They took a short - but smooth - detour to end up at the same destination as their colleague.
Natural Language Understanding
If you look at the example above, the LabVoice user offered two affirmative answers, “Sure” and “Yeah”, and LabVoice understood that. How does that happen?
That’s Natural Language Understanding, or NLU, in a nutshell. NLU is actually a sub-topic of Natural Language Processing, or NLP, which works with human languages to disseminate them into a machine-readable format.
NLU takes the various grammar rules and common sayings, phrases, and syntax to make interactions a more human-like experience. NLU helps LabVoice understand users’ intent, rather than requiring the exact wording or phrase to move forward in a process.
Leveraging NLU is extremely important from a user experience perspective. Would it be the end of the world for a user to reply “Yes” instead of “Sure” or “Yeah”? Of course not. But does it cut down on frustrations and distractions as users move through processes? Absolutely. Think about a compound search function, and all of the various names you could call one particular compound: its elemental name, its internal alphanumeric code, its location, etc.
Voice Authentication and Improved Speech Recognition
As a LabVoice user, not only are you assigned an email and password login to access the platform, you also have to voice-enroll, saying “Hey LabVoice” a few times for LabVoice to remember who you are. This helps us greatly from a security standpoint, as then we can ensure only registered users are able to access the platform, but like NLU, actually helps improve the user experience and data integrity.
When you use LabVoice, the platform can recognize not only who is speaking to it but improve its recognition of what that unique user is saying as well. LabVoice can account for idiosyncrasies like accents and phrases and thus improve its speech recognition for that user over some time.
In a scientific setting, this is important not only to eliminate distractions and frustrations like with NLU but also to improve the quality of the data that LabVoice captures. By utilizing its understanding of the user’s speech patterns, LabVoice is primed to handle tricky scientific vocabularies and user-specific terms, especially compared to everyday voice assistants like Siri. That means less time correcting errors (human and machine), transcribing data, and cleaning up messy data sets.
Process Performance Analytics
When taken collectively, each of these subtopics creates the foundational layer of laboratory continuous improvement. When connected with a lab’s infrastructure - defined here by its process, people, software, and instruments - LabVoice’s AI capabilities help users in several different ways. As the data is collected at the laboratory process with AI, the data is clean and FAIR, allowing future queries and analytics to be performed in an efficient and accurate manner.
Consider the scenario with instrument usage. As an HPLC is used throughout the analytical procedures, AI can observe retention time shifts. With this data and information about the last calibration, AI can predict equipment failures and suggest corrective measures such as column maintenance or replacement.
The same is true for consumable inventory levels. As reagents are used and stocks are depleted, LabVoice, and remind users when it’s time to purchase more for inventory.
We hope this helps you better understand how LabVoice utilizes artificial intelligence. If you have questions, don’t hesitate to talk to us - use the chatbot nearby or email us at firstname.lastname@example.org