Assistance animals are accepted as an invaluable tool for the sight-impaired, deaf, and those with other disabilities. However, some research indicates that dogs, or even cats, can be beneficial for individuals with a wide array of disabilities. Evidence suggests trained dogs can, in some cases, sniff out tumorous cells, predict the onset of seizures or hypoglycemic episodes. 
Service animals currently fall into two primary groups; response animals and alert animals. Response animals are characterized by being trained to act specific way to guide the human to the correct action. For example, a service animal might alert a sight-impaired person to the presence of steps, or a curb.
Alert animals are characterized by alerting their human counterparts to the onset of disabling events, such as a seizure or hypoglycemic episode. An alert dog might bark, or otherwise alert the human to the oncoming event. Alert animals are intended to give the human ample time to prepare for the hypoglycemic episode or seizure. Being alerted by the service animal allows the diabetic to treat himself with insulin, or other medications. It allows those with seizures time to take medication or lie down to protect themselves from the traumas that can occur from their surroundings during a seizure.
Response animals are traditionally used for the sight-impaired or hearing-impaired, while alert animals may provide life changing relief for those who suffer from diabetes or seizures.
Service animals for these cases are currently being drawn from three primary sources; Dog rescue centers, dogs specifically donated for the purpose of being trained, and so-called “career change” dogs, who were previously service animals, but were no longer required in their roles. Interestingly, most current research indicates that gender, breed, and age have no affect on the effectiveness of the service animal. More important is the bond between animal and owner. There seems to be a correlation between the strength of the bond between animal and human, and the likelihood that the animal will accurately predict seizures or hypoglycemic episodes. 
Currently, studies have been unable to determine how alert animals determine the onset of these episodes. There are several competing theories as to what causes an animal to alert. Studies suggest a dogs heightened sense of smell, which can detect minute changes in scent, may alert a service animal as to when a hypoglycemic episode or seizure is about to occur. Another theory suggests it may be subtle changes in the person’s behavior in the minutes before onset. Dogs may perceive these changes more easily than the person. Additionally, it has been contended that increased heart rate may be perceptible by service animals, alerting them to the onset of a seizure or hypoglycemic episode.
An interesting side effect of alert animals is a decrease in the frequency in seizures. Patients who had an effective alert animal at their side showed a significant decrease in the number of seizures they had in a 6 month period. Additionally, these patients were less depressed and more able to live a fulfilling life compared to those who did not have effective service animals. Researchers suggest that the patient feeling some control over their condition (by knowing when a seizure will occur) decreases depression and anxiety, which, in turn, decreases the frequency of seizures. 
Research proving the effectiveness of trained service animals for seizure and hypoglycemic detection is still in its infancy, and is mostly reliant on anecdotal evidence. The studies conducted, however, have shown a proclivity for both dogs and cats to determine these episodes.
State of Current Research
Stephen W. Brown and Laura H. Goldstein (2011) conducted an index observation in which trained alert dogs were paired with humans suffering from tonic-clonic seizures. This study led the researchers to further refine training techniques. The study, while not conclusively proving the effectiveness of service animals to predict seizures, did provide evidence supporting the idea that service animals at least decrease frequency of seizures. Comparing the baseline study of frequency of seizures to the final 12 week check up, there was a 43% decrease in the overall number of seizures. 9 of the 10 participants showed a 34% or greater reduction, and 4 of the 10 participants showed a reduction greater than 50%. Only one patient showed no signs of improvement. For the patients that did improve, the most significant drop occurred during the first 4 weeks of training. This study also highlighted the potential dangers involved in training service animals. The study reports 36 cases of service animals suffering serious health affects, with three resulting in the death of the animal. These are primarily caused by untrained animals reacting adversely to seizures and causing injury. However, in cases with trained service animals, these reactions were not seen. 
Deborah J. Dalziel, Basim M. Uthman, Susan P. McGorray, and Roger L. Reep (2003) conducted a literature review and preliminary study to determine the efficacy of service alert dogs. The study also found associated costs, training centers, behaviors that indicated alerting and factors that determined a dogs effectiveness. The preliminary study consisted of a qualitative questionnaire for epilepsy patients. 63 patients were involved, with 29 owning dogs. Of those, 9 reported their dogs reacted to seizures, while 3 reported their dogs predicted the onset of seizures. The study found that of 110 service dog training facilities in the United States, only 15 trained seizure assist dogs. The study also found the exorbitant costs involved in training a service animal, from $6-24,000. This cost is often not offset by insurance, making service alert animals unaffordable to many. The study also suggested age, breed, and gender were not factors in determining the effectiveness of a service animal. The behaviors used by the animal, however, could vary widely. A primary concern with alert animals is training the human to recognize the signals the alert animal is giving. These signals could include barking, pawing, whining, or staring. Patients indicated they needed to differentiate between alerting behaviors and general attention seeking behaviors. This study is concedes that it is too small in scale to be considered truly conclusive, but a few things seem to be certain; those with partial complex seizures are likely to be alerted, those with migraines are more likely to be alerted, and those who experience dizziness, nausea, lip smacking or changes in breathing are more likely to be alerted. Issues that need to be addressed are the lack of standardized training, and the preventative high cost associated with service alert animals. 
Strong and Brown (2002) conducted a study using a specialized training protocol, specifically to gauge the efficacy of service alert animals. The training was designed to give useful alert to the owner, and prevent injury to the service animal. The study based its methods on the theory that dogs alert based on subtle changes in the humans behavior in the minutes before the onset of an episode. In the study, Tonic-Clonic seizure frequency was recorded on a monthly basis. The baseline average was 13.8 seizures. During the 12 week training period, the average number of seizures dropped to 9.7. The second and third 12 week periods also showed decreases in the frequency of seizures, to 8.8 and 8.5, respectively. This means the service dogs decreased the number of seizures by, on average, 5.3. Participants stated the ability to predict seizures allowed them to engage in more activities, and fear the condition less. The decrease in frequency of seizures may be attributed to the increase in confidence created by the service alert dogs. 
Deborah L. Wells, Shaun W. Lawson, and A. Niroshan Siriwardena conducted a study to investigate the ability of service alert dogs to respond to hypoglycemia in patients with type 1 diabetes. In the study, 212 dog owners with medically diagnosed type 1 diabetes completed a questionnaire to gather information on the dogs’ responses to their hypoglycemic episodes. 65.1% responded their pet responded in some way to one or more of their hypoglycemic episodes, and 31.9% reported their animals responding to 11 events or more. As with other studies, age, breed and gender showed no correlation to likelihood of response. The study found most animals alerted before the participant realized he or she was hypoglycemic. Alerting behaving included barking, licking, nuzzling, jumping, or staring. A small portion of the animals showed fear-like responses, such as trembling, running away, or hyperventilating. The study also determined most people were not adept at recognizing their own symptoms for an oncoming hypoglycemic episode. Their pets, however, were better able to recognize the minute changes in behavior that would occur before a hypoglycemic episode. 
Areas for Investigation
There is still much research to be done in order to effectively prove the efficacy of service alert animals. At this time, most evidence is anecdotal, or from small scale studies. Larger studies need to be done in order to for service alert animals to become commonplace.
Researchers have yet to reach consensus as to which cues animals pick up on in detecting these episodes. Additional research is needed to isolate which cues result in accurate detection. A study is needed to determine the cause for dogs to alert, whether its subtle changes in human behavior, scent-oriented, or by sensing a change in heart rate. Identifying the cause for alert could prove valuable for training future service alert animals.
^ a b Deborah L. Wells, Shaun W. Lawson, and A. Niroshan Siriwardena (2008). Canine Responses to Hypoglycemia in Patients with Type 1 Diabetes. The Journal of Alternative and Complementary Medicine. Vol 14 No. 10 1235-1241.