Forensic Applications
Last year the author filed case reports with computer match statistics in over 75 criminal cases. The offenses include sexual assault, homicide, weapons or drug possession, bank robbery, and home invasion. Expert testimony was given in state, federal, military, and foreign courts. Most cases involved police or prosecutors needing a match statistic for arrest or court on DNA evidence where human review was inconclusive or unpersuasive. The DNA was usually crucial evidence, with the defendants typically convicted of their crimes. Several cases involved the defense assessing actual or post conviction innocence.

The courts have accepted computer DNA evidence interpretation. The Pennsylvania appellate Superior Court upheld the 2009 Commonwealth v. Kevin Foley homicide conviction and computer admissibility in a published precedential ruling.²³ In the Real IRA Massereene Barracks attack, which killed two unarmed British soldiers, the Northern Ireland court admitted computer methodology into evidence and used the DNA match statistic in its ruling.²⁴ These legal precedents are based on extensive scientific validations of the probabilistic genotype method^15,16 and regulatory approval.²⁵ Computer interpretation is not novel, just a useful implementation based on established mathematics and science.

Highly informative computer-inferred evidence genotypes can help investigators find criminals and missing people. However, some government DNA databases institutionalize the low information yield of human interpretation methods. One crime lab study showed that out of fifty DNA mixtures, human review did not produce a match statistic on half the evidence items, whereas the computer succeeded every time. Other crime lab studies show that even when human review of mixture data does yield a match statistic, the computer’s numbers are (on average) about a million times greater. More informative genotypes translate into a more powerful investigative database.

Investigative DNA databases of information-rich genotypes were used for identifying victim remains in the World Trade Center disaster.²⁶ The statistical computer inferred probabilistic genotypes from victim remains evidence, and from missing person kinship and personal effects data; the investigative database then compared these genotypes to form matches having LR statistics.²⁷ Probabilistic genotypes are written into the ANSI/NIST forensic data exchange standards (sect. 18.020-18.021),²⁸ and are specifically allowed by SWGDAM interpretation guidelines (par. 3.2.2).¹³

Path Forward
An information age demands information. We expect a Google search to return thorough, objective, and informative results, using the best available probability computer model methods. Our human minds ask questions, and we rely on the computer to calculate the best answers. Whether cracking a code, diagnosing disease, piloting a plane, or working on Wall Street, our lives and livelihoods depend on computer thought. Apprehending criminals through forensic intelligence is no exception—we want the most informative computers working 24/7 to provide protection.

DNA laboratories are now bringing computers on board to extend their forensic examiners’ analytic capability. A scientist can organize evidence and frame forensic questions; robots and computers can then automate the mechanics. A forensic scientist can incorporate informative DNA match statistics from complex mixture calculations into their case reports, and provide testimony in court. Experts excel at human activities, while computers are better calculators. Even before their crime labs deploy computer interpretation, police investigators and trial attorneys can rely on the private sector to deliver computer processing, case reports, and expert witness services.