Caring for a young infant, although a potentially rewarding means of producing a labor force for chores and minor home repairs, can be a trying ordeal for both new and experienced parents. The peaks and valleys of parental experience can leave a caregiver both exhilarated and agonizingly frustrated during a single hour of childcare, let alone the first few months. It is not an uncommon experience for a parent to rapidly alternate between extreme states of emotional arousal, one minute gazing down at their sleeping baby with seemingly limitless feelings of joy and love, and the next panicking at the perception that it has been too long since their baby’s last breath.
Babies, especially stupid ones, require near constant attention during the first several weeks of life, and that’s if it is going well. There is no user manual for the care of the newborn human that could possibly describe every situation and how to effectively respond to it in each individual child. A trial and error approach is always necessary to some degree, and it tends to result in a lot of sleepless nights, with many parents finding themselves more exhausted than they ever dreamed possible. So it shouldn’t be surprising that parents are a particularly vulnerable population when it comes to the marketing claims of bogus technology aimed at making their lives even the slightest bit easier.
Introducing the Infant Cries Translator
With that in mind, consider yet another example of technology that violates the “too good to be true” dictum: the Infant Cries Translator. Developed by a team of Taiwanese researchers from the National Taiwan University Hospital Yunlin and credulously reported on by the media over the past two weeks, this “new” development in childcare comes in the form of a smartphone application and carries the claim of being able to translate what the cries of a young infant actually mean. It’s nonsense. Complete and utter nonsense.
Why the scare quotes around new? Well, because despite the implications of the researchers and press, this exact technology already exists and previous versions have even had their own rounds of uncritical reporting in both 2009 and 2013. The only differences are the developer and the price. Though equally worthless, the more recent smartphone applications are at least considerably cheaper than the older baby translating devices that had to be plugged in near a wailing infant.
So what does this latest iteration of baby translator claim to be able to do? According to the developers it can tell the difference between four different cries within just 15 seconds of recording. The potential results include hunger, having a wet diaper, sleepiness, and pain. Conveniently for the developers, as I’ll explain shortly, the application is most accurate when used on infants that are less than 2 weeks of age. They in fact claim a whopping 92% accuracy in this age group. As a child ages, their cries are increasingly more difficult to translate, resulting in an accuracy of 85% in a 1-2 month old and 77% in a 4 month old. They say to not even waste your time once a child is more than 6 months old.
An exhaustive review of the extensive literature on baby cry interpretation
What are these accuracy claims based on? The researchers, Chang Chuan-yu and pediatrician Dr. Chen Si-da, first collected and analyzed 200,000 cries from 100 newborns for their “subtle differences in acoustics.” Then a miracle occurs, after which they collected feedback from users in order to come up with those numbers. That’s it.
There is no study, or even an explanation that I can find, of how they turned the 200,000 cries into an algorithm for the application. And despite the complete lack of actual data looking into accuracy, not that it would even be possible, or comparing infants of varying ages or locations, the app requires input of your baby’s date of birth and nationality. It’s hard to imagine that their 100 “subjects” were representative of too many regions of the world. Of course it doesn’t really matter because it’s all nonsense.
One of the earlier versions of cry interpretation technology, the Cry Translator from Biloop Technologies, claims to be able to tell a worried parent within 3 seconds if their baby is hungry, sleepy, annoyed, stressed, or bored. Guess which one I am. But at least this company provides a link to a clinical trial.
The study, designed and run by one of the product’s developers, was published in Spanish. But thanks to the wonders of Google, I was able to translate this clinical trial into English and to appreciate the full glory of its methodology. Dr. Portugal Ramirez, a “leading pediatrician and specialist” collected data from 104 infants over a roughly 2 year period. Data collection involved using the translator to determine why the baby was crying, which was then confirmed by a physician.
If the intervention suggested by the device led to calming of the infant’s crying, it was counted as a positive result, which occurred 96% of the time. Dr. Portugal Ramirez, based on the astounding success of his device and a profound misunderstanding of infant neurodevelopment, goes on to imply that regular use of the technology will increase a baby’s IQ. So what’s the problem with this study?
The validation technique employed in this study is comical. First off, the physician operating the device is unblinded and so just happens to have been involved in its development. He used it to translate each episode of crying and followed the recommended interventions. If the baby responded to the intervention by calming, it is counted as an accurate “diagnosis.”
The five possible reasons for crying and their respective interventions are as follows:
– Give the bottle
– Check thirst
– Go to a quiet place
– Gentle rocking in arms
– Talking soft
– Rocking in the fetal position
– Check room temperature
– Check diaper
– Check gas
– Burp the infant
– Check for constipation
– Check temperature
– Stimulate the baby
– Gentle rocking
– Talking softly to baby
Notice anything? Each category involves a physical interaction with the crying child. Simply picking a crying infant up is often effective at calming them. So is feeding them, regardless of whether or not it is time for them to eat because sucking is a natural calming reflex in the young infants and an effective distraction even as they age. So if a child’s cries were interpreted as being related to boredom, which is a rather silly concept (don’t anthropomorphize babies, they hate that), the interventions from any of the possible category would still likely work.
And just what constitutes the successful calming of a subject isn’t defined at all in the study. Was it complete cessation of all crying or perhaps just a noticeable decrease in intensity? This is a rather large researcher degree of freedom. It’s almost as if this study was designed to be positive in order to support the marketing of the device.
The younger a baby is, the more likely such simple physical interventions will be effective. Gentle rocking will simply not be as helpful when dealing with a 6-month-old brute as it is in a newborn. This is why the misguided (or malicious) researchers behind the latest cry interpretation app found such amazing results in infants that were less than 2-weeks-old.
Why do babies cry?
The elephant in the room when it comes to these devices and applications is the question of how did the developers come up with their algorithms in the first place. That is if we assume that they aren’t just Magic 8-Ball equivalents that spit out random responses. Unfortunately, there really is no reliable means of determining why a young infant is crying in the vast majority of circumstances, hence the trial and error approach I mentioned previously.
Obviously babies will cry if hungry or uncomfortable, but they also cry for no discernible reason. We assume that they have to mean something but many episodes are likely just the result of a developing brain being stimulated. The fact of the matter is that we don’t really know why babies cry most of the time and it doesn’t matter how much experience or training an observer has, how big of a database researchers work with, or how many “subtle acoustic variations” they discover.
It’s true that parents will sometimes claim that they can differentiate different types of crying in their own child. They will describe a hunger cry, a scared cry, or a cry that they interpret as one caused by some kind of discomfort such as gas or constipation. But these anecdotes are largely based on confirmation bias. We forget the misses and remember the hits because it helps us feel better when we have something to blame or focus on during times of uncertainty. Unfortunately this universal human approach to uncertainty that has led to many false beliefs and subsequent risky interventions.
There are of course extreme examples where the nature of a baby’s cry is unusual and perhaps more indicative of a specific cause, such as a craniofacial abnormality, genetic syndrome, or significant pain. But for the most part the nature of a child’s cry isn’t a very helpful clue for either parents or physicians. This technology only worsens the confusion.
Conclusion: Worry + pseudosolution = profit!
Baby translating technologies are implausible and unproven. The only thing these applications and devices do if used as recommended is translate a parent’s worry into profits for disreputable businessmen and gullible developers. Any pediatricians involved in the development or marketing of these products should be ashamed of themselves.
Thankfully, I don’t think much harm would come from their use. It isn’t as if they are diagnosing crying children with demonic possession or some fictional medical diagnosis like baby adrenal insufficiency. And I could see this being a funny gag gift meant purely for entertainment purposes.
I can’t help but imagine how much easier my job would be if this kind of technology actually worked, if baby cries actually were attempts to communicate specific complaints or desires. It truly would be a boon to parents and medical professionals caring for young infants. This is the impetus for parents of disabled older children to seek out other demonstrably ineffective but considerably more harmful technologies like facilitated communication.