What are the biggest obstacles standing in the way of your success as a veterinary business owner, leader or support team member?
- Chronic doctor shortages and a lack of qualified, trained team members?
- A sense that you are spread too thin between your roles as owner, leader, and healthcare worker?
- Ongoing, debilitating, intraoffice squabbles?
- A shortage of people that can mentor your newer team members and help them overcome their insecurities?
- Increased corporate competition and a fight for new clients?
- A growing cohort of clients demanding more for less?
Now, what if I told you that Artificial Intelligence (A.I.) is poised to fast track you past many, if not all, of these hurdles?
Artificial intelligence will radically change the way you work. In some ways, the changes will come so rapidly that you’ll feel a bit likeAlice in Wonderland – not so much passing through change, but falling headlong into it .
What is Artificial Intelligence?
Artificial intelligence is simulated human intelligence. It is machines that have been programmed to think and learn like humans. In conventional computing, problem solving is linear. A finite set of data is entered into a computer and the machine calculates an answer based on a series of if /then statements. Artificial computing on the other hand considers an enormous amount of information all at once, arrives at assumptions about that data, then weighs those assumptions against a vast pool of additional information. During this sizing up process, A.I. is looking for patterns, determining if what it thinks matches past correct answers and what is known to be true.
Here’s an example. Let’s say that a computer is assigned the job of transcribing what your doctor has handwritten into a medical chart and then comprehending what she means.
Since you have experience reading the scrawl in a handwritten medical chart, you know that this is not an easy assignment. Not only is it hard to determine what the words are, the sentences are often so attenuated, it can be a challenge to figure out what a doctor means. How can a computer decode this?
Well, let’s first consider how you do it. When you read a handwritten medical chart, you start by reading the words you can identify. In the example we are using, even the first word is a challenge. It starts with something resembling a P, its next letter could be an L, the third letter is anyone’s guess, and the fourth and final letter looks like an N. P-L-Something-N. That’s not too hard. Because this isn’t your first day on the job, you quickly conclude that the word must be Plan.
In the same manner as above, you work your way through the rest of the words in the sentence and finally resolve that the sentence reads: Plan: Fast patient for the next 12 hours and slowly reintroduce food. Send home w/d.
Though the words you have read have many letters that you can’t identify and the sentence is somewhat idiomatic (What is a ‘fast patient’ and what is a w/d?), you have quickly, probably without consciously thinking, concluded that this scribble means that the pet shouldn’t be given any food until tomorrow morning and then should be fed a prescription diet manufactured by Hill’s Pet Nutrition. You were able to do this because you have knowledge of the English language and know that PL-N can be no other word but PLAN. You know this not just because PLEN, PLIN, PLON, and PLUN do not exist in our language, but also because you’ve read dozens of this doctor’s charts before and know that she always lists a PLAN in her charts as part of her S.O.A.P. Additionally, because you understand medical terminology, have read the text preceeding the Plan, and know that the patient presented with diarrhea, you rightly conclude that Fast patient doesn’t mean that the patient is racing through the office, but that food should be withheld until morning. You also understand that w/d isn’t a reference to the auto body lubricant WD-40, but the pet food. Giving the pet w/d food for its upset stomach would make sense, where spraying its joints with W/D 40 lubricant would not.
This has been a relatively easy task for you, but imagine how hard it would be to get a computer to do this. There is no linear if/then statement that can assist a computer with knowing all the myriad ways the letters P L A or N can be drawn and no straightforward direction that can instruct the computer to understand that ‘Send home w/d’ implies that the client should be sold a bag of food before she leaves. In order to teach the computer to think on this level, we have to find a way for it to compute comprehensively. This is done with three essential ingredients.
Fast, robust computing power
Like our human minds, artificially intelligent computers have to consider the problem they endeavor to solve from many angles, give each of its considerations weight, and then match its solutions against a database of known answers and other information. To do this, A.I. computers require a lot of computing strength, now possible with the computing power of our modern-day chips.
An abundance of data
In order to ‘think’ comprehensively, A.I. requires a vast amount of data from which it can deduce trends, patterns, and context. To understand the importance of data in decision making, look out the window now. Do you see an oak tree? What about a pine tree? What about a rose bush? As a child, you were taught that tall stiff stalks with leaves on them were called trees. Later, after a few examples, you learned that tall stiff stalks with needles on them were trees, but qualified as pine ones, and that shorter, squatter stalks with leaves weren’t trees, but bushes. Since then, you have seen tens-of-thousands of examples of trees and bushes. Every one that you can remember instantly enters into your decision-making process when you are trying to identify a regular tree, a pine one, or a bush.
Today, I can fly you to the other side of the world and point to any number of stalks with leaves on them, and though you have never seen any of these individual specimens before, you can rightly identify all of them as either bushes or trees. That’s because you have a wealth of previous examples to draw from and because you are using a kind of reasoning power that concludes, ‘This stalk I’m looking at looks so much like the stalks I saw back at home, it must be a tree.’
A.I. is built to learn like you. As the amount of data from which a computer sources information rises, so does its ability to see patterns, similarities or differences, and reach a solution that makes sense. A.I. can follow linear pathways of thinking: if you see a tall stalk with leaves on top, then it is a tree, but it can also be taught to consider the entirety of the plant: its leaves, its color, where it is, how it is planted, what’s hanging from it, what the bark looks like, the size of its thorns, etc., and then compare that data against the million other stalks in its database. From there it can conclude that the specimen is a tree or a bush.
Another example is how your phone predicts what you’re about to type next. Every time you type a word in a sentence, your phone instantly matches your sentence fragment against millions of other fragments in its database. It considers the context of what you are talking about and what other people have chosen to write in similar situations, then your phone makes three guesses at the word you are about to write next and lists these suggestions at the bottom of your screen.
This kind of deductive reasoning is only going to get better. At some point in the future, your phone’s A.I. feature will also take into consideration how you personally talk and think as part of its calculations. Soon, it will be impossible for the people receiving your text to distinguish whether you or your phone authored it.
Advanced algorithms
The last thing that A.I. needs to be successful are algorithms, complex computer coding comprehensive enough to consider the many data points needed for human-like decision making. Interestingly, this code is currently not just written by people. As we have advanced in our work with A.I., we have built computers that can write code for themselves, grade their own success, and then rewrite the code to improve their own performance. It’s the equivalent of you taking a test at school, earning a B, then getting up in front of the classroom and reteaching the lesson to yourself so that next time you are tested, you get an A.
A day in the life of an A.I. practice
Now that you have a better understanding of how A.I. works, let’s think about what we are trying to accomplish on a day-to-day basis and see how A.I. can be of help.
The intake nurse is a computer
It’s 2034. A client has a sick cat. She logs onto your website and begins a discussion with an A.I.-powered chatbot-nurse version of you, but this isn’t the text bubble we know today. This chatbot has more of the appearance of a FaceTime video. It looks like you, sounds like you, phrases things like you. It has your weird sense of humor and, because it remembers the exchanges that it has had with this client in the past, asks things like, “So how are your kids doing? And I hope your other cat, Lilac, is feeling better.”
Companies like Delphi.ai invite clients to use a variety of ways to upload writing and speaking samples to improve a clone’s likeness.
This image of you is called a digital clone. It’s constructed using the power of A.I. and has been programmed using your life’s history of emails, text messages, any videos or messages you ever recorded, even results of the movies you have watched or searched for, all the videos you’ve looked at and articles that you have read. Every scrap of digital information that has ever been collected on you has been potentially sourced, including a digital form of all of the previous medical charts notes you have made, so A.I. knows how you think (and talk) medically. Lastly, your clone designer has digitally taped hundreds of hours of you talking about your thoughts, feelings, your past, your dreams, your aspirations and so forth. She has gathered millions of pieces of data in an attempt to create the most authentic, and, I might add, the best version of you, because, just like everyone who’s ever had their picture taken will tell you, no one wants a digital clone to be authentic, they want it to be ideal.
The nurse-chatbot version of you takes a complete history from the client. Because time is not money when it comes to employing your digital clone (or not as much money), this history can be exhaustive, ensuring that you are getting the most information possible on this case, a boon to clients that feel rushed during the history process and worry that you don’t fully understand their concerns.
When the history is through, your digital self instantly parses through all the information: current history, past history, past signalment, and past diagnostics, and then uploads a concise, short summary of all that she knows. Think of this as that major/minor problem list you currently have in your patient charts. Then, based on the urgency of the signs and the client’s schedule, the A.I. tool schedules the patient for a visit. This visit can be scheduled as either a virtual or in person exam, your preference, and like the good nurse that your digital clone is, she ensures the client knows about all about your hospital’s latest parasite testing and prevention promotions.
Upon arrival, the human version of you greets the client and carries your digital clone into the room on a tablet or opens her up on one of the exam room monitors. Human-you does a TPR while your clone is on hand to record all that transpires. When this part of the exam is complete, human-you retreats to fetch the human-doctor, while digital-you remains in the room to keep the client company, to discuss the pet’s ongoing progress, and so forth.
Finally, both human-you and the human-doctor enter the room. The doctor does the examination while human-you holds the pet, prepares injections, gathers supplies, etc. In the meantime, digital-you, using A.I. voice recognition technology, records all that happens and enters it into the patient chart. When the doctor is finished, she outlines a plan, but most likely this plan has already been sketched out by your clone, more concisely, comprehensively, and accurately than the doctor herself.
Your digital self assists, but also collaborates with the doctor. Remember your digital clone has access to millions of bits of data on this breed, at this age, weight and geographical location. Your clone knows that weather in the area has been wetter than normal and that wetter weather increases the likelihood that your patient may have Giardia. It knows the likelihood that this case before you will require more extensive care and whether you have or will have the capacity in your hospital to care for it. It also has access to laboratory data showing that Parvovirus diagnoses are threefold what they normally are and that your patient’s breed is two-times more likely to contract the disease . The pet may be outfitted with a wearable device that has been recording a stream of continuous metrics on the patient. Your digital self instantly scans this data for patterns, a chore so big that it might be impossible for a human to undertake, but for your digital clone, it is done in an eye blink.
Having broadly outlined the medical treatment plan to the client, the doctor leaves the room to see the next patient. The digital version of you reviews the granular points of the treatment plan, potentially helps the client apply for financing, and then calls the human version of you into the room to start diagnostics and treatment. At the end of the appointment, the digital clone checks that human-you has filled medication orders accurately, confirms the dosages to eliminate any chance of error, schedules the client’s next appointment, and then takes payment for services.
Though the client leaves the human version of you at the practice, your digital-you continues to communicate with her long after she has left. Your digital clone can spend hours with her, as she is simultaneously doing with dozens of your other clients, improving patient outcomes, client satisfaction, and reducing the amount of strain on you, your doctors and your coworkers.
Also helping internal communication and conflict mediation
Missing that soft touch. A legitimate case for pushback against A.I. in the vet space?
When I share my thoughts on A.I. with veterinary audiences, a significant number of attendees sniff, “But where is the human touch? What about the empathy?”
Never mind that some of these folks’ demeanors are so doleful that clients (and coworkers!) haven’t had a ‘touch’ or whiff of empathy in months, now these gloomy glums are concerned about exhibitions of kindness!
But I’ll agree with you. Interacting with a bot for only a short while can be a fast track to frustration. Forget that most of the answers that bots provide are designed to be generalized; forget that they often serve as an expensive phone trees; just knowing that our questions and concerns have been given to a machine to address, and not a person, signals dismissal. Yes, the machine is recording our information, but it is not appreciating the depth of what we are experiencing. That’s important to us humans.
Let’s use this client’s description of why she has come to the veterinarian as an example. “Since last night, he woke me up 6 times to go to the bathroom and once he couldn’t hold it. The poor guy. You could see how ashamed he was for having had an accident in the house.”
When the client tells us this story, she is not just conveying that the dog had 6 episodes of diarrhea. She wants us to understand the symptoms in context of her and her dog’s feelings because she believes that how they are experiencing the symptoms is every bit as important as the symptoms themselves. She is hoping that if words fail her as she describes what’s going on, that our comprehension of her experience of the symptoms will fill in the blanks; that we’ll understand what she fully means because we fully empathize with how she feels.
Still, having said all of that, how many times have we found ourselves so shorthanded, so overworked, or surrounded by so many untrained individuals, that we’re incapable of doing any of the above?
If you have dismissed A.I. as a solution to your veterinary business’s growth because it can’t feel like a human, then you’ve ignored the full scope of the problems that you are trying to resolve. Yes, empathy, is vital to healthcare, but so is answering the phone, having a way to make an appointment, to write up a chart, to refill a medication, and to follow through. A.I. can do all of these things; do them all with speed, accuracy, intelligence and no breaks.
If A.I. is not going to be your next veterinary nurse, she’s going to be on the payroll in some capacity and my sense is that she will be one of the most reliable and affordable hires in your hospital’s history.
But A.I won’t just help with clients and patients; it will also help you deal with your coworkers.
Let’s say that tempers have been smoldering between you and the practice owner. You need to talk to her about a recent medical case that went awry, but you’re dreading the discussion. In the past, when you’ve had these discussions with her, she’s always been reactive and you’ve been reactive back. In the end, you haven’t resolved anything, only worsened the animosity and mistrust you have for one another.
But with the use of digital clones, the human versions of you arrive to your meeting, sit back, and watch the virtual versions hash it out. A conversation that might otherwise have devolved into acrimony if humans had been running the show is maturely, fairly, and amicably discussed by your clones. In the end, your digital selves achieve an accord and reveal a path to peace that you and your human colleague can mirror in the future.
When Does All This Start and What are My Next Steps?
It has started. The following companies are all employing A.I. in the veterinary space. Here’s a little bit more of who they are and how they are using A.I. to improve.
Processes
Our sponsor, Hill’s Pet Nutrition, employs artificial intelligence at its Tonganoxie facility. The factory features advanced automation, robotics, and digital quality assessments to optimize production processes. Hill’s also partners with Vet-AI to enhance veterinary care using AI-powered solutions.
Diagnostics, assessments, plans, and mentorship
IDEXX Laboratories leverages artificial intelligence to detect complex diseases more accurately and quickly. For instance, the IDEXX SediVue Dx analyzer uses AI to examine urine samples and then to provide rapid and reliable results. Its Dx Cellular Analyzer employs AI to automate the examination of blood and cytology samples and its AI-powered practice management systems interpret diagnostic results and recommend next steps for treatment. This helps veterinarians make more informed clinical decisions and improves the overall efficiency of veterinary practices. Moreover, the service constantly improves because every sample submitted adds to the amount of data that A.I. can reference and learn from.
Efficiency and service
Digitail offers an all-in-one cloud-based practice management software that’s integrated with AI tools. Their AI assistant, Tails AI, provides voice-to-text transcription, patient history summaries, diagnostic assistance, and personalized client communication, improving efficiency and reducing manual record-keeping time. Another company, Goldie Vet, uses automatic speech recognition (ASR) and natural language processing (NLP) to transcribe and analyze conversations between veterinarians and pet owners, creating comprehensive and accurate medical records.
Patient wellbeing
Sylvester AI has developed an AI tool called Tably that interprets cats’ facial cues to assess pain levels. This tool helps in post-procedure care and ongoing health monitoring by providing accurate pain assessments based on facial indicators.
What’s next for me and my practice?
The investment, funding, computer power and data resources required for A.I.’s use in the vet space is considerable and, unless something dramatically changes in the next few years, a deterrent to individual practices trying to implement it from the ground up. That is to say, no single doctor is going to build her own A.I. platform the same way she built her own website or her own app. It’s more likely that the power of A.I. will be available to practices first through existing affiliations with companies like those just mentioned. As A.I. use increases, more companies will flood the market in the same way that options for computing, computer storage, and communication have emerged and evolved with the rise of the internet.
In the meantime, you and your practice can think about the data you will require to make the best use of A.I. If you haven’t yet considered a cloud-based software system, now’s the time to take a look. Only cloud-computing software, not your clunky server-based stuff, will be able to grow and change with the needs of A.I. Ensure that the apps you are using integrate with your software so that you maximize the data sets available to you and ensure that your team members are using the software to its full potential, filling out all the information fields and working through, not around, your software to complete your work.
It’s also okay to dream big. Though we must still manage the major stressors of our profession – staff shortages and conflict, demanding clients, high caseloads – the opportunity for relief has arguably turned the corner. You and your team are about to be part of a new era in veterinary medicine where safety, care, and personal achievement all reach new heights.
Will Your Digital Self Get Paid?
Delphi.ai is already capable of creating clones that look, sound and express themselves just like you. Will there be several versions of you working in veterinary hospitals across America for which the real you will be paid?
Are you the vet that every client loves? Are you a tech with whiz-bang client communication skills? Are you a CSR with an astounding track record of turning phone shoppers into appointments? You might want to create a clone and make a fortune renting your digital self to other hospitals in need of your stellar abilities.
Today it is possible to have a digital version of yourself created, replete with your voice, your manner of speaking, the clothes you like to wear, your inflection, and even your sense of humor. It is also possible for designers to give you blonde highlights or an entirely new style, a makeover with the version of your face that existed 15 years ago, and the tattoos you always wanted but your husband never allowed you to buy. Then, you can sell this image of yourself, h ave it embed it in any veterinary website in the world, and make it earn money for you, 24 hours a day, 7 days a week for the rest of your life.
It’s not science fiction. It’s happening today and it’s a possibility for you to consider as you dream of that day when you live out early retirement under an umbrella by the sea. Here’s how it will work:
- You’ll amass a database of how you talk, how you perform in front of clients, and what you think. The best way to do this would be for you to approach your life much like a social media influencer would. You’ll write online content, shoot videos, create posts of things that interest you, collect samples of your medical writings, and demonstrate how you do the things that you do best: reassure clients that wait time won’t be too much longer; break down medical costs for clients into separate appointments to make bills more manageable, wrap instrument packs, talk to clients about the value of spaying their 8 year old dog, etc.
- You’ll work with an existing company like Delphi.ai to upload this information and to create and host your digital clone. Other partners might include companies like IBM Watson Health, Google Health, or universities interested in learning how to scale this concept.
- An interface will be created that allows these large companies to connect your digital clone to your company’s smaller database or the database of the company for whom your clone will work. This will allow your clone to search the information of every client that calls or lands on the site and to create a personalized experience for each.
- Your clone will be embedded into websites and phone trees and it will interact with as many clients as the platform can manage.
- You’ll be compensated a number of ways. You may choose to sell your digital self out right for a one-time fee, sell it as a subscription service where you collect money on a monthly or yearly basis, collect for the total amount of time that your clone spends interacting with clients, or charge every time you convert a lead.
- Continue to tweak your clone’s performance by taping additional original material or adjusting the programming for what you have already.